var zE=Object.defineProperty;var RE=(Is,Hr,Nn)=>Hr in Is?zE(Is,Hr,{enumerable:!0,configurable:!0,writable:!0,value:Nn}):Is[Hr]=Nn;var re=(Is,Hr,Nn)=>RE(Is,typeof Hr!="symbol"?Hr+"":Hr,Nn);(function(){"use strict";const Is=new Map,Hr=[],Nn=(e,r,t)=>{if(r&&typeof r.init=="function"&&typeof r.createInferenceSessionHandler=="function"){const s=Is.get(e);if(s===void 0)Is.set(e,{backend:r,priority:t});else{if(s.priority>t)return;if(s.priority===t&&s.backend!==r)throw new Error(`cannot register backend "${e}" using priority ${t}`)}if(t>=0){const o=Hr.indexOf(e);o!==-1&&Hr.splice(o,1);for(let n=0;n<Hr.length;n++)if(Is.get(Hr[n]).priority<=t){Hr.splice(n,0,e);return}Hr.push(e)}return}throw new TypeError("not a valid backend")},lx=async e=>{const r=Is.get(e);if(!r)return"backend not found.";if(r.initialized)return r.backend;if(r.aborted)return r.error;{const t=!!r.initPromise;try{return t||(r.initPromise=r.backend.init(e)),await r.initPromise,r.initialized=!0,r.backend}catch(s){return t||(r.error=`${s}`,r.aborted=!0),r.error}finally{delete r.initPromise}}},ux=async e=>{const r=e.executionProviders||[],t=r.map(l=>typeof l=="string"?l:l.name),s=t.length===0?Hr:t;let o;const n=[],a=new Set;for(const l of s){const c=await lx(l);typeof c=="string"?n.push({name:l,err:c}):(o||(o=c),o===c&&a.add(l))}if(!o)throw new Error(`no available backend found. ERR: ${n.map(l=>`[${l.name}] ${l.err}`).join(", ")}`);for(const{name:l,err:c}of n)t.includes(l)&&console.warn(`removing requested execution provider "${l}" from session options because it is not available: ${c}`);const i=r.filter(l=>a.has(typeof l=="string"?l:l.name));return[o,new Proxy(e,{get:(l,c)=>c==="executionProviders"?i:Reflect.get(l,c)})]},cx="1.23.0";let Zc="warning";const Br={wasm:{},webgl:{},webgpu:{},versions:{common:cx},set logLevel(e){if(e!==void 0){if(typeof e!="string"||["verbose","info","warning","error","fatal"].indexOf(e)===-1)throw new Error(`Unsupported logging level: ${e}`);Zc=e}},get logLevel(){return Zc}};Object.defineProperty(Br,"logLevel",{enumerable:!0});const dx=Br,px=(e,r)=>{const t=typeof document<"u"?document.createElement("canvas"):new OffscreenCanvas(1,1);t.width=e.dims[3],t.height=e.dims[2];const s=t.getContext("2d");if(s!=null){let o,n;(r==null?void 0:r.tensorLayout)!==void 0&&r.tensorLayout==="NHWC"?(o=e.dims[2],n=e.dims[3]):(o=e.dims[3],n=e.dims[2]);const a=(r==null?void 0:r.format)!==void 0?r.format:"RGB",i=r==null?void 0:r.norm;let l,c;i===void 0||i.mean===void 0?l=[255,255,255,255]:typeof i.mean=="number"?l=[i.mean,i.mean,i.mean,i.mean]:(l=[i.mean[0],i.mean[1],i.mean[2],0],i.mean[3]!==void 0&&(l[3]=i.mean[3])),i===void 0||i.bias===void 0?c=[0,0,0,0]:typeof i.bias=="number"?c=[i.bias,i.bias,i.bias,i.bias]:(c=[i.bias[0],i.bias[1],i.bias[2],0],i.bias[3]!==void 0&&(c[3]=i.bias[3]));const p=n*o;let d=0,u=p,_=p*2,f=-1;a==="RGBA"?(d=0,u=p,_=p*2,f=p*3):a==="RGB"?(d=0,u=p,_=p*2):a==="RBG"&&(d=0,_=p,u=p*2);for(let b=0;b<n;b++)for(let A=0;A<o;A++){const g=(e.data[d++]-c[0])*l[0],y=(e.data[u++]-c[1])*l[1],C=(e.data[_++]-c[2])*l[2],x=f===-1?255:(e.data[f++]-c[3])*l[3];s.fillStyle="rgba("+g+","+y+","+C+","+x+")",s.fillRect(A,b,1,1)}if("toDataURL"in t)return t.toDataURL();throw new Error("toDataURL is not supported")}else throw new Error("Can not access image data")},hx=(e,r)=>{const t=typeof document<"u"?document.createElement("canvas").getContext("2d"):new OffscreenCanvas(1,1).getContext("2d");let s;if(t!=null){let o,n,a;(r==null?void 0:r.tensorLayout)!==void 0&&r.tensorLayout==="NHWC"?(o=e.dims[2],n=e.dims[1],a=e.dims[3]):(o=e.dims[3],n=e.dims[2],a=e.dims[1]);const i=r!==void 0&&r.format!==void 0?r.format:"RGB",l=r==null?void 0:r.norm;let c,p;l===void 0||l.mean===void 0?c=[255,255,255,255]:typeof l.mean=="number"?c=[l.mean,l.mean,l.mean,l.mean]:(c=[l.mean[0],l.mean[1],l.mean[2],255],l.mean[3]!==void 0&&(c[3]=l.mean[3])),l===void 0||l.bias===void 0?p=[0,0,0,0]:typeof l.bias=="number"?p=[l.bias,l.bias,l.bias,l.bias]:(p=[l.bias[0],l.bias[1],l.bias[2],0],l.bias[3]!==void 0&&(p[3]=l.bias[3]));const d=n*o;if(r!==void 0&&(r.format!==void 0&&a===4&&r.format!=="RGBA"||a===3&&r.format!=="RGB"&&r.format!=="BGR"))throw new Error("Tensor format doesn't match input tensor dims");const u=4;let _=0,f=1,b=2,A=3,g=0,y=d,C=d*2,x=-1;i==="RGBA"?(g=0,y=d,C=d*2,x=d*3):i==="RGB"?(g=0,y=d,C=d*2):i==="RBG"&&(g=0,C=d,y=d*2),s=t.createImageData(o,n);for(let M=0;M<n*o;_+=u,f+=u,b+=u,A+=u,M++)s.data[_]=(e.data[g++]-p[0])*c[0],s.data[f]=(e.data[y++]-p[1])*c[1],s.data[b]=(e.data[C++]-p[2])*c[2],s.data[A]=x===-1?255:(e.data[x++]-p[3])*c[3]}else throw new Error("Can not access image data");return s},Ti=(e,r)=>{if(e===void 0)throw new Error("Image buffer must be defined");if(r.height===void 0||r.width===void 0)throw new Error("Image height and width must be defined");if(r.tensorLayout==="NHWC")throw new Error("NHWC Tensor layout is not supported yet");const{height:t,width:s}=r,o=r.norm??{mean:255,bias:0};let n,a;typeof o.mean=="number"?n=[o.mean,o.mean,o.mean,o.mean]:n=[o.mean[0],o.mean[1],o.mean[2],o.mean[3]??255],typeof o.bias=="number"?a=[o.bias,o.bias,o.bias,o.bias]:a=[o.bias[0],o.bias[1],o.bias[2],o.bias[3]??0];const i=r.format!==void 0?r.format:"RGBA",l=r.tensorFormat!==void 0&&r.tensorFormat!==void 0?r.tensorFormat:"RGB",c=t*s,p=l==="RGBA"?new Float32Array(c*4):new Float32Array(c*3);let d=4,u=0,_=1,f=2,b=3,A=0,g=c,y=c*2,C=-1;i==="RGB"&&(d=3,u=0,_=1,f=2,b=-1),l==="RGBA"?C=c*3:l==="RBG"?(A=0,y=c,g=c*2):l==="BGR"&&(y=0,g=c,A=c*2);for(let M=0;M<c;M++,u+=d,f+=d,_+=d,b+=d)p[A++]=(e[u]+a[0])/n[0],p[g++]=(e[_]+a[1])/n[1],p[y++]=(e[f]+a[2])/n[2],C!==-1&&b!==-1&&(p[C++]=(e[b]+a[3])/n[3]);return l==="RGBA"?new is("float32",p,[1,4,t,s]):new is("float32",p,[1,3,t,s])},mx=async(e,r)=>{const t=typeof HTMLImageElement<"u"&&e instanceof HTMLImageElement,s=typeof ImageData<"u"&&e instanceof ImageData,o=typeof ImageBitmap<"u"&&e instanceof ImageBitmap,n=typeof e=="string";let a,i=r??{};const l=()=>{if(typeof document<"u")return document.createElement("canvas");if(typeof OffscreenCanvas<"u")return new OffscreenCanvas(1,1);throw new Error("Canvas is not supported")},c=p=>typeof HTMLCanvasElement<"u"&&p instanceof HTMLCanvasElement||p instanceof OffscreenCanvas?p.getContext("2d"):null;if(t){const p=l();p.width=e.width,p.height=e.height;const d=c(p);if(d!=null){let u=e.height,_=e.width;if(r!==void 0&&r.resizedHeight!==void 0&&r.resizedWidth!==void 0&&(u=r.resizedHeight,_=r.resizedWidth),r!==void 0){if(i=r,r.tensorFormat!==void 0)throw new Error("Image input config format must be RGBA for HTMLImageElement");i.tensorFormat="RGBA",i.height=u,i.width=_}else i.tensorFormat="RGBA",i.height=u,i.width=_;d.drawImage(e,0,0),a=d.getImageData(0,0,_,u).data}else throw new Error("Can not access image data")}else if(s){let p,d;if(r!==void 0&&r.resizedWidth!==void 0&&r.resizedHeight!==void 0?(p=r.resizedHeight,d=r.resizedWidth):(p=e.height,d=e.width),r!==void 0&&(i=r),i.format="RGBA",i.height=p,i.width=d,r!==void 0){const u=l();u.width=d,u.height=p;const _=c(u);if(_!=null)_.putImageData(e,0,0),a=_.getImageData(0,0,d,p).data;else throw new Error("Can not access image data")}else a=e.data}else if(o){if(r===void 0)throw new Error("Please provide image config with format for Imagebitmap");const p=l();p.width=e.width,p.height=e.height;const d=c(p);if(d!=null){const u=e.height,_=e.width;return d.drawImage(e,0,0,_,u),a=d.getImageData(0,0,_,u).data,i.height=u,i.width=_,Ti(a,i)}else throw new Error("Can not access image data")}else{if(n)return new Promise((p,d)=>{const u=l(),_=c(u);if(!e||!_)return d();const f=new Image;f.crossOrigin="Anonymous",f.src=e,f.onload=()=>{u.width=f.width,u.height=f.height,_.drawImage(f,0,0,u.width,u.height);const b=_.getImageData(0,0,u.width,u.height);i.height=u.height,i.width=u.width,p(Ti(b.data,i))}});throw new Error("Input data provided is not supported - aborted tensor creation")}if(a!==void 0)return Ti(a,i);throw new Error("Input data provided is not supported - aborted tensor creation")},fx=(e,r)=>{const{width:t,height:s,download:o,dispose:n}=r,a=[1,s,t,4];return new is({location:"texture",type:"float32",texture:e,dims:a,download:o,dispose:n})},_x=(e,r)=>{const{dataType:t,dims:s,download:o,dispose:n}=r;return new is({location:"gpu-buffer",type:t??"float32",gpuBuffer:e,dims:s,download:o,dispose:n})},gx=(e,r)=>{const{dataType:t,dims:s,download:o,dispose:n}=r;return new is({location:"ml-tensor",type:t??"float32",mlTensor:e,dims:s,download:o,dispose:n})},wx=(e,r,t)=>new is({location:"cpu-pinned",type:e,data:r,dims:t??[r.length]}),jn=new Map([["float32",Float32Array],["uint8",Uint8Array],["int8",Int8Array],["uint16",Uint16Array],["int16",Int16Array],["int32",Int32Array],["bool",Uint8Array],["float64",Float64Array],["uint32",Uint32Array],["int4",Uint8Array],["uint4",Uint8Array]]),ra=new Map([[Float32Array,"float32"],[Uint8Array,"uint8"],[Int8Array,"int8"],[Uint16Array,"uint16"],[Int16Array,"int16"],[Int32Array,"int32"],[Float64Array,"float64"],[Uint32Array,"uint32"]]);let ed=!1;const Mx=()=>{if(!ed){ed=!0;const e=typeof BigInt64Array<"u"&&BigInt64Array.from,r=typeof BigUint64Array<"u"&&BigUint64Array.from,t=globalThis.Float16Array,s=typeof t<"u"&&t.from;e&&(jn.set("int64",BigInt64Array),ra.set(BigInt64Array,"int64")),r&&(jn.set("uint64",BigUint64Array),ra.set(BigUint64Array,"uint64")),s?(jn.set("float16",t),ra.set(t,"float16")):jn.set("float16",Uint16Array)}},bx=e=>{let r=1;for(let t=0;t<e.length;t++){const s=e[t];if(typeof s!="number"||!Number.isSafeInteger(s))throw new TypeError(`dims[${t}] must be an integer, got: ${s}`);if(s<0)throw new RangeError(`dims[${t}] must be a non-negative integer, got: ${s}`);r*=s}return r},yx=(e,r)=>{switch(e.location){case"cpu":return new is(e.type,e.data,r);case"cpu-pinned":return new is({location:"cpu-pinned",data:e.data,type:e.type,dims:r});case"texture":return new is({location:"texture",texture:e.texture,type:e.type,dims:r});case"gpu-buffer":return new is({location:"gpu-buffer",gpuBuffer:e.gpuBuffer,type:e.type,dims:r});case"ml-tensor":return new is({location:"ml-tensor",mlTensor:e.mlTensor,type:e.type,dims:r});default:throw new Error(`tensorReshape: tensor location ${e.location} is not supported`)}};let is=class{constructor(r,t,s){Mx();let o,n;if(typeof r=="object"&&"location"in r)switch(this.dataLocation=r.location,o=r.type,n=r.dims,r.location){case"cpu-pinned":{const i=jn.get(o);if(!i)throw new TypeError(`unsupported type "${o}" to create tensor from pinned buffer`);if(!(r.data instanceof i))throw new TypeError(`buffer should be of type ${i.name}`);this.cpuData=r.data;break}case"texture":{if(o!=="float32")throw new TypeError(`unsupported type "${o}" to create tensor from texture`);this.gpuTextureData=r.texture,this.downloader=r.download,this.disposer=r.dispose;break}case"gpu-buffer":{if(o!=="float32"&&o!=="float16"&&o!=="int32"&&o!=="int64"&&o!=="uint32"&&o!=="uint8"&&o!=="bool"&&o!=="uint4"&&o!=="int4")throw new TypeError(`unsupported type "${o}" to create tensor from gpu buffer`);this.gpuBufferData=r.gpuBuffer,this.downloader=r.download,this.disposer=r.dispose;break}case"ml-tensor":{if(o!=="float32"&&o!=="float16"&&o!=="int32"&&o!=="int64"&&o!=="uint32"&&o!=="uint64"&&o!=="int8"&&o!=="uint8"&&o!=="bool"&&o!=="uint4"&&o!=="int4")throw new TypeError(`unsupported type "${o}" to create tensor from MLTensor`);this.mlTensorData=r.mlTensor,this.downloader=r.download,this.disposer=r.dispose;break}default:throw new Error(`Tensor constructor: unsupported location '${this.dataLocation}'`)}else{let i,l;if(typeof r=="string")if(o=r,l=s,r==="string"){if(!Array.isArray(t))throw new TypeError("A string tensor's data must be a string array.");i=t}else{const c=jn.get(r);if(c===void 0)throw new TypeError(`Unsupported tensor type: ${r}.`);if(Array.isArray(t)){if(r==="float16"&&c===Uint16Array||r==="uint4"||r==="int4")throw new TypeError(`Creating a ${r} tensor from number array is not supported. Please use ${c.name} as data.`);r==="uint64"||r==="int64"?i=c.from(t,BigInt):i=c.from(t)}else if(t instanceof c)i=t;else if(t instanceof Uint8ClampedArray)if(r==="uint8")i=Uint8Array.from(t);else throw new TypeError("A Uint8ClampedArray tensor's data must be type of uint8");else if(r==="float16"&&t instanceof Uint16Array&&c!==Uint16Array)i=new globalThis.Float16Array(t.buffer,t.byteOffset,t.length);else throw new TypeError(`A ${o} tensor's data must be type of ${c}`)}else if(l=t,Array.isArray(r)){if(r.length===0)throw new TypeError("Tensor type cannot be inferred from an empty array.");const c=typeof r[0];if(c==="string")o="string",i=r;else if(c==="boolean")o="bool",i=Uint8Array.from(r);else throw new TypeError(`Invalid element type of data array: ${c}.`)}else if(r instanceof Uint8ClampedArray)o="uint8",i=Uint8Array.from(r);else{const c=ra.get(r.constructor);if(c===void 0)throw new TypeError(`Unsupported type for tensor data: ${r.constructor}.`);o=c,i=r}if(l===void 0)l=[i.length];else if(!Array.isArray(l))throw new TypeError("A tensor's dims must be a number array");n=l,this.cpuData=i,this.dataLocation="cpu"}const a=bx(n);if(this.cpuData&&a!==this.cpuData.length&&!((o==="uint4"||o==="int4")&&Math.ceil(a/2)===this.cpuData.length))throw new Error(`Tensor's size(${a}) does not match data length(${this.cpuData.length}).`);this.type=o,this.dims=n,this.size=a}static async fromImage(r,t){return mx(r,t)}static fromTexture(r,t){return fx(r,t)}static fromGpuBuffer(r,t){return _x(r,t)}static fromMLTensor(r,t){return gx(r,t)}static fromPinnedBuffer(r,t,s){return wx(r,t,s)}toDataURL(r){return px(this,r)}toImageData(r){return hx(this,r)}get data(){if(this.ensureValid(),!this.cpuData)throw new Error("The data is not on CPU. Use `getData()` to download GPU data to CPU, or use `texture` or `gpuBuffer` property to access the GPU data directly.");return this.cpuData}get location(){return this.dataLocation}get texture(){if(this.ensureValid(),!this.gpuTextureData)throw new Error("The data is not stored as a WebGL texture.");return this.gpuTextureData}get gpuBuffer(){if(this.ensureValid(),!this.gpuBufferData)throw new Error("The data is not stored as a WebGPU buffer.");return this.gpuBufferData}get mlTensor(){if(this.ensureValid(),!this.mlTensorData)throw new Error("The data is not stored as a WebNN MLTensor.");return this.mlTensorData}async getData(r){switch(this.ensureValid(),this.dataLocation){case"cpu":case"cpu-pinned":return this.data;case"texture":case"gpu-buffer":case"ml-tensor":{if(!this.downloader)throw new Error("The current tensor is not created with a specified data downloader.");if(this.isDownloading)throw new Error("The current tensor is being downloaded.");try{this.isDownloading=!0;const t=await this.downloader();return this.downloader=void 0,this.dataLocation="cpu",this.cpuData=t,r&&this.disposer&&(this.disposer(),this.disposer=void 0),t}finally{this.isDownloading=!1}}default:throw new Error(`cannot get data from location: ${this.dataLocation}`)}}dispose(){if(this.isDownloading)throw new Error("The current tensor is being downloaded.");this.disposer&&(this.disposer(),this.disposer=void 0),this.cpuData=void 0,this.gpuTextureData=void 0,this.gpuBufferData=void 0,this.mlTensorData=void 0,this.downloader=void 0,this.isDownloading=void 0,this.dataLocation="none"}ensureValid(){if(this.dataLocation==="none")throw new Error("The tensor is disposed.")}reshape(r){if(this.ensureValid(),this.downloader||this.disposer)throw new Error("Cannot reshape a tensor that owns GPU resource.");return yx(this,r)}};const Vn=is,td=(e,r)=>{(typeof Br.trace>"u"?!Br.wasm.trace:!Br.trace)||console.timeStamp(`${e}::ORT::${r}`)},rd=(e,r)=>{var o;const t=((o=new Error().stack)==null?void 0:o.split(/\r\n|\r|\n/g))||[];let s=!1;for(let n=0;n<t.length;n++){if(s&&!t[n].includes("TRACE_FUNC")){let a=`FUNC_${e}::${t[n].trim().split(" ")[1]}`;r&&(a+=`::${r}`),td("CPU",a);return}t[n].includes("TRACE_FUNC")&&(s=!0)}},Ei=e=>{(typeof Br.trace>"u"?!Br.wasm.trace:!Br.trace)||rd("BEGIN",e)},Pi=e=>{(typeof Br.trace>"u"?!Br.wasm.trace:!Br.trace)||rd("END",e)},Ci=e=>{(typeof Br.trace>"u"?!Br.wasm.trace:!Br.trace)||console.time(`ORT::${e}`)},Si=e=>{(typeof Br.trace>"u"?!Br.wasm.trace:!Br.trace)||console.timeEnd(`ORT::${e}`)};var vx=Object.freeze({__proto__:null,InferenceSession:class ox{constructor(r){this.handler=r}async run(r,t,s){Ei(),Ci("InferenceSession.run");const o={};let n={};if(typeof r!="object"||r===null||r instanceof Vn||Array.isArray(r))throw new TypeError("'feeds' must be an object that use input names as keys and OnnxValue as corresponding values.");let a=!0;if(typeof t=="object"){if(t===null)throw new TypeError("Unexpected argument[1]: cannot be null.");if(t instanceof Vn)throw new TypeError("'fetches' cannot be a Tensor");if(Array.isArray(t)){if(t.length===0)throw new TypeError("'fetches' cannot be an empty array.");a=!1;for(const c of t){if(typeof c!="string")throw new TypeError("'fetches' must be a string array or an object.");if(this.outputNames.indexOf(c)===-1)throw new RangeError(`'fetches' contains invalid output name: ${c}.`);o[c]=null}if(typeof s=="object"&&s!==null)n=s;else if(typeof s<"u")throw new TypeError("'options' must be an object.")}else{let c=!1;const p=Object.getOwnPropertyNames(t);for(const d of this.outputNames)if(p.indexOf(d)!==-1){const u=t[d];(u===null||u instanceof Vn)&&(c=!0,a=!1,o[d]=u)}if(c){if(typeof s=="object"&&s!==null)n=s;else if(typeof s<"u")throw new TypeError("'options' must be an object.")}else n=t}}else if(typeof t<"u")throw new TypeError("Unexpected argument[1]: must be 'fetches' or 'options'.");for(const c of this.inputNames)if(typeof r[c]>"u")throw new Error(`input '${c}' is missing in 'feeds'.`);if(a)for(const c of this.outputNames)o[c]=null;const i=await this.handler.run(r,o,n),l={};for(const c in i)if(Object.hasOwnProperty.call(i,c)){const p=i[c];p instanceof Vn?l[c]=p:l[c]=new Vn(p.type,p.data,p.dims)}return Si("InferenceSession.run"),Pi(),l}async release(){return this.handler.dispose()}static async create(r,t,s,o){Ei(),Ci("InferenceSession.create");let n,a={};if(typeof r=="string"){if(n=r,typeof t=="object"&&t!==null)a=t;else if(typeof t<"u")throw new TypeError("'options' must be an object.")}else if(r instanceof Uint8Array){if(n=r,typeof t=="object"&&t!==null)a=t;else if(typeof t<"u")throw new TypeError("'options' must be an object.")}else if(r instanceof ArrayBuffer||typeof SharedArrayBuffer<"u"&&r instanceof SharedArrayBuffer){const p=r;let d=0,u=r.byteLength;if(typeof t=="object"&&t!==null)a=t;else if(typeof t=="number"){if(d=t,!Number.isSafeInteger(d))throw new RangeError("'byteOffset' must be an integer.");if(d<0||d>=p.byteLength)throw new RangeError(`'byteOffset' is out of range [0, ${p.byteLength}).`);if(u=r.byteLength-d,typeof s=="number"){if(u=s,!Number.isSafeInteger(u))throw new RangeError("'byteLength' must be an integer.");if(u<=0||d+u>p.byteLength)throw new RangeError(`'byteLength' is out of range (0, ${p.byteLength-d}].`);if(typeof o=="object"&&o!==null)a=o;else if(typeof o<"u")throw new TypeError("'options' must be an object.")}else if(typeof s<"u")throw new TypeError("'byteLength' must be a number.")}else if(typeof t<"u")throw new TypeError("'options' must be an object.");n=new Uint8Array(p,d,u)}else throw new TypeError("Unexpected argument[0]: must be 'path' or 'buffer'.");const[i,l]=await ux(a),c=await i.createInferenceSessionHandler(n,l);return Si("InferenceSession.create"),Pi(),new ox(c)}startProfiling(){this.handler.startProfiling()}endProfiling(){this.handler.endProfiling()}get inputNames(){return this.handler.inputNames}get outputNames(){return this.handler.outputNames}get inputMetadata(){return this.handler.inputMetadata}get outputMetadata(){return this.handler.outputMetadata}},TRACE:td,TRACE_EVENT_BEGIN:Ci,TRACE_EVENT_END:Si,TRACE_FUNC_BEGIN:Ei,TRACE_FUNC_END:Pi,Tensor:Vn,env:dx,registerBackend:Nn});/*!
 * ONNX Runtime Web v1.22.0-dev.20250409-89f8206ba4
 * Copyright (c) Microsoft Corporation. All rights reserved.
 * Licensed under the MIT License.
 */var $i=Object.defineProperty,xx=Object.getOwnPropertyDescriptor,Tx=Object.getOwnPropertyNames,Ex=Object.prototype.hasOwnProperty,Px=(e=>typeof require<"u"?require:typeof Proxy<"u"?new Proxy(e,{get:(r,t)=>(typeof require<"u"?require:r)[t]}):e)(function(e){if(typeof require<"u")return require.apply(this,arguments);throw Error('Dynamic require of "'+e+'" is not supported')}),Ue=(e,r)=>()=>(e&&(r=e(e=0)),r),Un=(e,r)=>{for(var t in r)$i(e,t,{get:r[t],enumerable:!0})},Cx=(e,r,t,s)=>{if(r&&typeof r=="object"||typeof r=="function")for(let o of Tx(r))!Ex.call(e,o)&&o!==t&&$i(e,o,{get:()=>r[o],enumerable:!(s=xx(r,o))||s.enumerable});return e},co=e=>Cx($i({},"__esModule",{value:!0}),e),po,Xs,pn,sd,nd,od=Ue(()=>{po=new Map,Xs=[],pn=(e,r,t)=>{if(r&&typeof r.init=="function"&&typeof r.createInferenceSessionHandler=="function"){let s=po.get(e);if(s===void 0)po.set(e,{backend:r,priority:t});else{if(s.priority>t)return;if(s.priority===t&&s.backend!==r)throw new Error(`cannot register backend "${e}" using priority ${t}`)}if(t>=0){let o=Xs.indexOf(e);o!==-1&&Xs.splice(o,1);for(let n=0;n<Xs.length;n++)if(po.get(Xs[n]).priority<=t){Xs.splice(n,0,e);return}Xs.push(e)}return}throw new TypeError("not a valid backend")},sd=async e=>{let r=po.get(e);if(!r)return"backend not found.";if(r.initialized)return r.backend;if(r.aborted)return r.error;{let t=!!r.initPromise;try{return t||(r.initPromise=r.backend.init(e)),await r.initPromise,r.initialized=!0,r.backend}catch(s){return t||(r.error=`${s}`,r.aborted=!0),r.error}finally{delete r.initPromise}}},nd=async e=>{let r=e.executionProviders||[],t=r.map(l=>typeof l=="string"?l:l.name),s=t.length===0?Xs:t,o,n=[],a=new Set;for(let l of s){let c=await sd(l);typeof c=="string"?n.push({name:l,err:c}):(o||(o=c),o===c&&a.add(l))}if(!o)throw new Error(`no available backend found. ERR: ${n.map(l=>`[${l.name}] ${l.err}`).join(", ")}`);for(let{name:l,err:c}of n)t.includes(l)&&console.warn(`removing requested execution provider "${l}" from session options because it is not available: ${c}`);let i=r.filter(l=>a.has(typeof l=="string"?l:l.name));return[o,new Proxy(e,{get:(l,c)=>c==="executionProviders"?i:Reflect.get(l,c)})]}}),Sx=Ue(()=>{od()}),ad,$x=Ue(()=>{ad="1.22.0-dev.20250409-89f8206ba4"}),Ai,ls,id=Ue(()=>{$x(),Ai="warning",ls={wasm:{},webgl:{},webgpu:{},versions:{common:ad},set logLevel(e){if(e!==void 0){if(typeof e!="string"||["verbose","info","warning","error","fatal"].indexOf(e)===-1)throw new Error(`Unsupported logging level: ${e}`);Ai=e}},get logLevel(){return Ai}},Object.defineProperty(ls,"logLevel",{enumerable:!0})}),Jt,Ax=Ue(()=>{id(),Jt=ls}),ld,ud,kx=Ue(()=>{ld=(e,r)=>{let t=typeof document<"u"?document.createElement("canvas"):new OffscreenCanvas(1,1);t.width=e.dims[3],t.height=e.dims[2];let s=t.getContext("2d");if(s!=null){let o,n;(r==null?void 0:r.tensorLayout)!==void 0&&r.tensorLayout==="NHWC"?(o=e.dims[2],n=e.dims[3]):(o=e.dims[3],n=e.dims[2]);let a=(r==null?void 0:r.format)!==void 0?r.format:"RGB",i=r==null?void 0:r.norm,l,c;i===void 0||i.mean===void 0?l=[255,255,255,255]:typeof i.mean=="number"?l=[i.mean,i.mean,i.mean,i.mean]:(l=[i.mean[0],i.mean[1],i.mean[2],0],i.mean[3]!==void 0&&(l[3]=i.mean[3])),i===void 0||i.bias===void 0?c=[0,0,0,0]:typeof i.bias=="number"?c=[i.bias,i.bias,i.bias,i.bias]:(c=[i.bias[0],i.bias[1],i.bias[2],0],i.bias[3]!==void 0&&(c[3]=i.bias[3]));let p=n*o,d=0,u=p,_=p*2,f=-1;a==="RGBA"?(d=0,u=p,_=p*2,f=p*3):a==="RGB"?(d=0,u=p,_=p*2):a==="RBG"&&(d=0,_=p,u=p*2);for(let b=0;b<n;b++)for(let A=0;A<o;A++){let g=(e.data[d++]-c[0])*l[0],y=(e.data[u++]-c[1])*l[1],C=(e.data[_++]-c[2])*l[2],x=f===-1?255:(e.data[f++]-c[3])*l[3];s.fillStyle="rgba("+g+","+y+","+C+","+x+")",s.fillRect(A,b,1,1)}if("toDataURL"in t)return t.toDataURL();throw new Error("toDataURL is not supported")}else throw new Error("Can not access image data")},ud=(e,r)=>{let t=typeof document<"u"?document.createElement("canvas").getContext("2d"):new OffscreenCanvas(1,1).getContext("2d"),s;if(t!=null){let o,n,a;(r==null?void 0:r.tensorLayout)!==void 0&&r.tensorLayout==="NHWC"?(o=e.dims[2],n=e.dims[1],a=e.dims[3]):(o=e.dims[3],n=e.dims[2],a=e.dims[1]);let i=r!==void 0&&r.format!==void 0?r.format:"RGB",l=r==null?void 0:r.norm,c,p;l===void 0||l.mean===void 0?c=[255,255,255,255]:typeof l.mean=="number"?c=[l.mean,l.mean,l.mean,l.mean]:(c=[l.mean[0],l.mean[1],l.mean[2],255],l.mean[3]!==void 0&&(c[3]=l.mean[3])),l===void 0||l.bias===void 0?p=[0,0,0,0]:typeof l.bias=="number"?p=[l.bias,l.bias,l.bias,l.bias]:(p=[l.bias[0],l.bias[1],l.bias[2],0],l.bias[3]!==void 0&&(p[3]=l.bias[3]));let d=n*o;if(r!==void 0&&(r.format!==void 0&&a===4&&r.format!=="RGBA"||a===3&&r.format!=="RGB"&&r.format!=="BGR"))throw new Error("Tensor format doesn't match input tensor dims");let u=4,_=0,f=1,b=2,A=3,g=0,y=d,C=d*2,x=-1;i==="RGBA"?(g=0,y=d,C=d*2,x=d*3):i==="RGB"?(g=0,y=d,C=d*2):i==="RBG"&&(g=0,C=d,y=d*2),s=t.createImageData(o,n);for(let M=0;M<n*o;_+=u,f+=u,b+=u,A+=u,M++)s.data[_]=(e.data[g++]-p[0])*c[0],s.data[f]=(e.data[y++]-p[1])*c[1],s.data[b]=(e.data[C++]-p[2])*c[2],s.data[A]=x===-1?255:(e.data[x++]-p[3])*c[3]}else throw new Error("Can not access image data");return s}}),sa,cd,dd,pd,hd,md,Ix=Ue(()=>{Ii(),sa=(e,r)=>{if(e===void 0)throw new Error("Image buffer must be defined");if(r.height===void 0||r.width===void 0)throw new Error("Image height and width must be defined");if(r.tensorLayout==="NHWC")throw new Error("NHWC Tensor layout is not supported yet");let{height:t,width:s}=r,o=r.norm??{mean:255,bias:0},n,a;typeof o.mean=="number"?n=[o.mean,o.mean,o.mean,o.mean]:n=[o.mean[0],o.mean[1],o.mean[2],o.mean[3]??255],typeof o.bias=="number"?a=[o.bias,o.bias,o.bias,o.bias]:a=[o.bias[0],o.bias[1],o.bias[2],o.bias[3]??0];let i=r.format!==void 0?r.format:"RGBA",l=r.tensorFormat!==void 0&&r.tensorFormat!==void 0?r.tensorFormat:"RGB",c=t*s,p=l==="RGBA"?new Float32Array(c*4):new Float32Array(c*3),d=4,u=0,_=1,f=2,b=3,A=0,g=c,y=c*2,C=-1;i==="RGB"&&(d=3,u=0,_=1,f=2,b=-1),l==="RGBA"?C=c*3:l==="RBG"?(A=0,y=c,g=c*2):l==="BGR"&&(y=0,g=c,A=c*2);for(let x=0;x<c;x++,u+=d,f+=d,_+=d,b+=d)p[A++]=(e[u]+a[0])/n[0],p[g++]=(e[_]+a[1])/n[1],p[y++]=(e[f]+a[2])/n[2],C!==-1&&b!==-1&&(p[C++]=(e[b]+a[3])/n[3]);return l==="RGBA"?new es("float32",p,[1,4,t,s]):new es("float32",p,[1,3,t,s])},cd=async(e,r)=>{let t=typeof HTMLImageElement<"u"&&e instanceof HTMLImageElement,s=typeof ImageData<"u"&&e instanceof ImageData,o=typeof ImageBitmap<"u"&&e instanceof ImageBitmap,n=typeof e=="string",a,i=r??{},l=()=>{if(typeof document<"u")return document.createElement("canvas");if(typeof OffscreenCanvas<"u")return new OffscreenCanvas(1,1);throw new Error("Canvas is not supported")},c=p=>typeof HTMLCanvasElement<"u"&&p instanceof HTMLCanvasElement||p instanceof OffscreenCanvas?p.getContext("2d"):null;if(t){let p=l();p.width=e.width,p.height=e.height;let d=c(p);if(d!=null){let u=e.height,_=e.width;if(r!==void 0&&r.resizedHeight!==void 0&&r.resizedWidth!==void 0&&(u=r.resizedHeight,_=r.resizedWidth),r!==void 0){if(i=r,r.tensorFormat!==void 0)throw new Error("Image input config format must be RGBA for HTMLImageElement");i.tensorFormat="RGBA",i.height=u,i.width=_}else i.tensorFormat="RGBA",i.height=u,i.width=_;d.drawImage(e,0,0),a=d.getImageData(0,0,_,u).data}else throw new Error("Can not access image data")}else if(s){let p,d;if(r!==void 0&&r.resizedWidth!==void 0&&r.resizedHeight!==void 0?(p=r.resizedHeight,d=r.resizedWidth):(p=e.height,d=e.width),r!==void 0&&(i=r),i.format="RGBA",i.height=p,i.width=d,r!==void 0){let u=l();u.width=d,u.height=p;let _=c(u);if(_!=null)_.putImageData(e,0,0),a=_.getImageData(0,0,d,p).data;else throw new Error("Can not access image data")}else a=e.data}else if(o){if(r===void 0)throw new Error("Please provide image config with format for Imagebitmap");let p=l();p.width=e.width,p.height=e.height;let d=c(p);if(d!=null){let u=e.height,_=e.width;return d.drawImage(e,0,0,_,u),a=d.getImageData(0,0,_,u).data,i.height=u,i.width=_,sa(a,i)}else throw new Error("Can not access image data")}else{if(n)return new Promise((p,d)=>{let u=l(),_=c(u);if(!e||!_)return d();let f=new Image;f.crossOrigin="Anonymous",f.src=e,f.onload=()=>{u.width=f.width,u.height=f.height,_.drawImage(f,0,0,u.width,u.height);let b=_.getImageData(0,0,u.width,u.height);i.height=u.height,i.width=u.width,p(sa(b.data,i))}});throw new Error("Input data provided is not supported - aborted tensor creation")}if(a!==void 0)return sa(a,i);throw new Error("Input data provided is not supported - aborted tensor creation")},dd=(e,r)=>{let{width:t,height:s,download:o,dispose:n}=r,a=[1,s,t,4];return new es({location:"texture",type:"float32",texture:e,dims:a,download:o,dispose:n})},pd=(e,r)=>{let{dataType:t,dims:s,download:o,dispose:n}=r;return new es({location:"gpu-buffer",type:t??"float32",gpuBuffer:e,dims:s,download:o,dispose:n})},hd=(e,r)=>{let{dataType:t,dims:s,download:o,dispose:n}=r;return new es({location:"ml-tensor",type:t??"float32",mlTensor:e,dims:s,download:o,dispose:n})},md=(e,r,t)=>new es({location:"cpu-pinned",type:e,data:r,dims:t??[r.length]})}),hn,ho,ki,fd,Fx=Ue(()=>{hn=new Map([["float32",Float32Array],["uint8",Uint8Array],["int8",Int8Array],["uint16",Uint16Array],["int16",Int16Array],["int32",Int32Array],["bool",Uint8Array],["float64",Float64Array],["uint32",Uint32Array],["int4",Uint8Array],["uint4",Uint8Array]]),ho=new Map([[Float32Array,"float32"],[Uint8Array,"uint8"],[Int8Array,"int8"],[Uint16Array,"uint16"],[Int16Array,"int16"],[Int32Array,"int32"],[Float64Array,"float64"],[Uint32Array,"uint32"]]),ki=!1,fd=()=>{if(!ki){ki=!0;let e=typeof BigInt64Array<"u"&&BigInt64Array.from,r=typeof BigUint64Array<"u"&&BigUint64Array.from,t=globalThis.Float16Array,s=typeof t<"u"&&t.from;e&&(hn.set("int64",BigInt64Array),ho.set(BigInt64Array,"int64")),r&&(hn.set("uint64",BigUint64Array),ho.set(BigUint64Array,"uint64")),s?(hn.set("float16",t),ho.set(t,"float16")):hn.set("float16",Uint16Array)}}}),_d,gd,Ox=Ue(()=>{Ii(),_d=e=>{let r=1;for(let t=0;t<e.length;t++){let s=e[t];if(typeof s!="number"||!Number.isSafeInteger(s))throw new TypeError(`dims[${t}] must be an integer, got: ${s}`);if(s<0)throw new RangeError(`dims[${t}] must be a non-negative integer, got: ${s}`);r*=s}return r},gd=(e,r)=>{switch(e.location){case"cpu":return new es(e.type,e.data,r);case"cpu-pinned":return new es({location:"cpu-pinned",data:e.data,type:e.type,dims:r});case"texture":return new es({location:"texture",texture:e.texture,type:e.type,dims:r});case"gpu-buffer":return new es({location:"gpu-buffer",gpuBuffer:e.gpuBuffer,type:e.type,dims:r});case"ml-tensor":return new es({location:"ml-tensor",mlTensor:e.mlTensor,type:e.type,dims:r});default:throw new Error(`tensorReshape: tensor location ${e.location} is not supported`)}}}),es,Ii=Ue(()=>{kx(),Ix(),Fx(),Ox(),es=class{constructor(e,r,t){fd();let s,o;if(typeof e=="object"&&"location"in e)switch(this.dataLocation=e.location,s=e.type,o=e.dims,e.location){case"cpu-pinned":{let a=hn.get(s);if(!a)throw new TypeError(`unsupported type "${s}" to create tensor from pinned buffer`);if(!(e.data instanceof a))throw new TypeError(`buffer should be of type ${a.name}`);this.cpuData=e.data;break}case"texture":{if(s!=="float32")throw new TypeError(`unsupported type "${s}" to create tensor from texture`);this.gpuTextureData=e.texture,this.downloader=e.download,this.disposer=e.dispose;break}case"gpu-buffer":{if(s!=="float32"&&s!=="float16"&&s!=="int32"&&s!=="int64"&&s!=="uint32"&&s!=="uint8"&&s!=="bool"&&s!=="uint4"&&s!=="int4")throw new TypeError(`unsupported type "${s}" to create tensor from gpu buffer`);this.gpuBufferData=e.gpuBuffer,this.downloader=e.download,this.disposer=e.dispose;break}case"ml-tensor":{if(s!=="float32"&&s!=="float16"&&s!=="int32"&&s!=="int64"&&s!=="uint32"&&s!=="uint64"&&s!=="int8"&&s!=="uint8"&&s!=="bool"&&s!=="uint4"&&s!=="int4")throw new TypeError(`unsupported type "${s}" to create tensor from MLTensor`);this.mlTensorData=e.mlTensor,this.downloader=e.download,this.disposer=e.dispose;break}default:throw new Error(`Tensor constructor: unsupported location '${this.dataLocation}'`)}else{let a,i;if(typeof e=="string")if(s=e,i=t,e==="string"){if(!Array.isArray(r))throw new TypeError("A string tensor's data must be a string array.");a=r}else{let l=hn.get(e);if(l===void 0)throw new TypeError(`Unsupported tensor type: ${e}.`);if(Array.isArray(r)){if(e==="float16"&&l===Uint16Array||e==="uint4"||e==="int4")throw new TypeError(`Creating a ${e} tensor from number array is not supported. Please use ${l.name} as data.`);e==="uint64"||e==="int64"?a=l.from(r,BigInt):a=l.from(r)}else if(r instanceof l)a=r;else if(r instanceof Uint8ClampedArray)if(e==="uint8")a=Uint8Array.from(r);else throw new TypeError("A Uint8ClampedArray tensor's data must be type of uint8");else if(e==="float16"&&r instanceof Uint16Array&&l!==Uint16Array)a=new globalThis.Float16Array(r.buffer,r.byteOffset,r.length);else throw new TypeError(`A ${s} tensor's data must be type of ${l}`)}else if(i=r,Array.isArray(e)){if(e.length===0)throw new TypeError("Tensor type cannot be inferred from an empty array.");let l=typeof e[0];if(l==="string")s="string",a=e;else if(l==="boolean")s="bool",a=Uint8Array.from(e);else throw new TypeError(`Invalid element type of data array: ${l}.`)}else if(e instanceof Uint8ClampedArray)s="uint8",a=Uint8Array.from(e);else{let l=ho.get(e.constructor);if(l===void 0)throw new TypeError(`Unsupported type for tensor data: ${e.constructor}.`);s=l,a=e}if(i===void 0)i=[a.length];else if(!Array.isArray(i))throw new TypeError("A tensor's dims must be a number array");o=i,this.cpuData=a,this.dataLocation="cpu"}let n=_d(o);if(this.cpuData&&n!==this.cpuData.length&&!((s==="uint4"||s==="int4")&&Math.ceil(n/2)===this.cpuData.length))throw new Error(`Tensor's size(${n}) does not match data length(${this.cpuData.length}).`);this.type=s,this.dims=o,this.size=n}static async fromImage(e,r){return cd(e,r)}static fromTexture(e,r){return dd(e,r)}static fromGpuBuffer(e,r){return pd(e,r)}static fromMLTensor(e,r){return hd(e,r)}static fromPinnedBuffer(e,r,t){return md(e,r,t)}toDataURL(e){return ld(this,e)}toImageData(e){return ud(this,e)}get data(){if(this.ensureValid(),!this.cpuData)throw new Error("The data is not on CPU. Use `getData()` to download GPU data to CPU, or use `texture` or `gpuBuffer` property to access the GPU data directly.");return this.cpuData}get location(){return this.dataLocation}get texture(){if(this.ensureValid(),!this.gpuTextureData)throw new Error("The data is not stored as a WebGL texture.");return this.gpuTextureData}get gpuBuffer(){if(this.ensureValid(),!this.gpuBufferData)throw new Error("The data is not stored as a WebGPU buffer.");return this.gpuBufferData}get mlTensor(){if(this.ensureValid(),!this.mlTensorData)throw new Error("The data is not stored as a WebNN MLTensor.");return this.mlTensorData}async getData(e){switch(this.ensureValid(),this.dataLocation){case"cpu":case"cpu-pinned":return this.data;case"texture":case"gpu-buffer":case"ml-tensor":{if(!this.downloader)throw new Error("The current tensor is not created with a specified data downloader.");if(this.isDownloading)throw new Error("The current tensor is being downloaded.");try{this.isDownloading=!0;let r=await this.downloader();return this.downloader=void 0,this.dataLocation="cpu",this.cpuData=r,e&&this.disposer&&(this.disposer(),this.disposer=void 0),r}finally{this.isDownloading=!1}}default:throw new Error(`cannot get data from location: ${this.dataLocation}`)}}dispose(){if(this.isDownloading)throw new Error("The current tensor is being downloaded.");this.disposer&&(this.disposer(),this.disposer=void 0),this.cpuData=void 0,this.gpuTextureData=void 0,this.gpuBufferData=void 0,this.mlTensorData=void 0,this.downloader=void 0,this.isDownloading=void 0,this.dataLocation="none"}ensureValid(){if(this.dataLocation==="none")throw new Error("The tensor is disposed.")}reshape(e){if(this.ensureValid(),this.downloader||this.disposer)throw new Error("Cannot reshape a tensor that owns GPU resource.");return gd(this,e)}}}),ws,wd=Ue(()=>{Ii(),ws=es}),mo,Fi,Ms,us,Md=Ue(()=>{id(),mo=(e,r)=>{(typeof ls.trace>"u"?!ls.wasm.trace:!ls.trace)||console.timeStamp(`${e}::ORT::${r}`)},Fi=(e,r)=>{var o;let t=((o=new Error().stack)==null?void 0:o.split(/\r\n|\r|\n/g))||[],s=!1;for(let n=0;n<t.length;n++){if(s&&!t[n].includes("TRACE_FUNC")){let a=`FUNC_${e}::${t[n].trim().split(" ")[1]}`;r&&(a+=`::${r}`),mo("CPU",a);return}t[n].includes("TRACE_FUNC")&&(s=!0)}},Ms=e=>{(typeof ls.trace>"u"?!ls.wasm.trace:!ls.trace)||Fi("BEGIN",e)},us=e=>{(typeof ls.trace>"u"?!ls.wasm.trace:!ls.trace)||Fi("END",e)}}),bd,Dx=Ue(()=>{od(),wd(),Md(),bd=class ax{constructor(r){this.handler=r}async run(r,t,s){Ms();let o={},n={};if(typeof r!="object"||r===null||r instanceof ws||Array.isArray(r))throw new TypeError("'feeds' must be an object that use input names as keys and OnnxValue as corresponding values.");let a=!0;if(typeof t=="object"){if(t===null)throw new TypeError("Unexpected argument[1]: cannot be null.");if(t instanceof ws)throw new TypeError("'fetches' cannot be a Tensor");if(Array.isArray(t)){if(t.length===0)throw new TypeError("'fetches' cannot be an empty array.");a=!1;for(let c of t){if(typeof c!="string")throw new TypeError("'fetches' must be a string array or an object.");if(this.outputNames.indexOf(c)===-1)throw new RangeError(`'fetches' contains invalid output name: ${c}.`);o[c]=null}if(typeof s=="object"&&s!==null)n=s;else if(typeof s<"u")throw new TypeError("'options' must be an object.")}else{let c=!1,p=Object.getOwnPropertyNames(t);for(let d of this.outputNames)if(p.indexOf(d)!==-1){let u=t[d];(u===null||u instanceof ws)&&(c=!0,a=!1,o[d]=u)}if(c){if(typeof s=="object"&&s!==null)n=s;else if(typeof s<"u")throw new TypeError("'options' must be an object.")}else n=t}}else if(typeof t<"u")throw new TypeError("Unexpected argument[1]: must be 'fetches' or 'options'.");for(let c of this.inputNames)if(typeof r[c]>"u")throw new Error(`input '${c}' is missing in 'feeds'.`);if(a)for(let c of this.outputNames)o[c]=null;let i=await this.handler.run(r,o,n),l={};for(let c in i)if(Object.hasOwnProperty.call(i,c)){let p=i[c];p instanceof ws?l[c]=p:l[c]=new ws(p.type,p.data,p.dims)}return us(),l}async release(){return this.handler.dispose()}static async create(r,t,s,o){Ms();let n,a={};if(typeof r=="string"){if(n=r,typeof t=="object"&&t!==null)a=t;else if(typeof t<"u")throw new TypeError("'options' must be an object.")}else if(r instanceof Uint8Array){if(n=r,typeof t=="object"&&t!==null)a=t;else if(typeof t<"u")throw new TypeError("'options' must be an object.")}else if(r instanceof ArrayBuffer||typeof SharedArrayBuffer<"u"&&r instanceof SharedArrayBuffer){let p=r,d=0,u=r.byteLength;if(typeof t=="object"&&t!==null)a=t;else if(typeof t=="number"){if(d=t,!Number.isSafeInteger(d))throw new RangeError("'byteOffset' must be an integer.");if(d<0||d>=p.byteLength)throw new RangeError(`'byteOffset' is out of range [0, ${p.byteLength}).`);if(u=r.byteLength-d,typeof s=="number"){if(u=s,!Number.isSafeInteger(u))throw new RangeError("'byteLength' must be an integer.");if(u<=0||d+u>p.byteLength)throw new RangeError(`'byteLength' is out of range (0, ${p.byteLength-d}].`);if(typeof o=="object"&&o!==null)a=o;else if(typeof o<"u")throw new TypeError("'options' must be an object.")}else if(typeof s<"u")throw new TypeError("'byteLength' must be a number.")}else if(typeof t<"u")throw new TypeError("'options' must be an object.");n=new Uint8Array(p,d,u)}else throw new TypeError("Unexpected argument[0]: must be 'path' or 'buffer'.");let[i,l]=await nd(a),c=await i.createInferenceSessionHandler(n,l);return us(),new ax(c)}startProfiling(){this.handler.startProfiling()}endProfiling(){this.handler.endProfiling()}get inputNames(){return this.handler.inputNames}get outputNames(){return this.handler.outputNames}get inputMetadata(){return this.handler.inputMetadata}get outputMetadata(){return this.handler.outputMetadata}}}),Oi,Lx=Ue(()=>{Dx(),Oi=bd}),zx=Ue(()=>{}),Rx=Ue(()=>{}),Bx=Ue(()=>{}),Nx=Ue(()=>{}),yd={};Un(yd,{InferenceSession:()=>Oi,TRACE:()=>mo,TRACE_FUNC_BEGIN:()=>Ms,TRACE_FUNC_END:()=>us,Tensor:()=>ws,env:()=>Jt,registerBackend:()=>pn});var bs=Ue(()=>{Sx(),Ax(),Lx(),wd(),zx(),Rx(),Md(),Bx(),Nx()}),Di=Ue(()=>{}),vd={};Un(vd,{default:()=>xd});var Li,zi,xd,jx=Ue(()=>{var e;sw(),mn(),Gi(),Li="ort-wasm-proxy-worker",zi=((e=globalThis.self)==null?void 0:e.name)===Li,zi&&(self.onmessage=r=>{let{type:t,in:s}=r.data;try{switch(t){case"init-wasm":qi(s.wasm).then(()=>{iu(s).then(()=>{postMessage({type:t})},o=>{postMessage({type:t,err:o})})},o=>{postMessage({type:t,err:o})});break;case"init-ep":{let{epName:o,env:n}=s;lu(n,o).then(()=>{postMessage({type:t})},a=>{postMessage({type:t,err:a})});break}case"copy-from":{let{buffer:o}=s,n=va(o);postMessage({type:t,out:n});break}case"create":{let{model:o,options:n}=s;cu(o,n).then(a=>{postMessage({type:t,out:a})},a=>{postMessage({type:t,err:a})});break}case"release":du(s),postMessage({type:t});break;case"run":{let{sessionId:o,inputIndices:n,inputs:a,outputIndices:i,options:l}=s;hu(o,n,a,i,new Array(i.length).fill(null),l).then(c=>{c.some(p=>p[3]!=="cpu")?postMessage({type:t,err:"Proxy does not support non-cpu tensor location."}):postMessage({type:t,out:c},fu([...a,...c]))},c=>{postMessage({type:t,err:c})});break}case"end-profiling":mu(s),postMessage({type:t});break;default:}}catch(o){postMessage({type:t,err:o})}}),xd=zi?null:r=>new Worker(r??ts,{type:"module",name:Li})}),Td={};Un(Td,{default:()=>Ed});var Ri,Bi,Ed,Pd,Vx=Ue(()=>{var e,r;Bi=(Ri=self.location.href,async function(t={}){var Jo;var s,o,n=t,a=new Promise((h,E)=>{s=h,o=E}),i=typeof window=="object",l=typeof WorkerGlobalScope<"u",c=l&&((Jo=self.name)==null?void 0:Jo.startsWith("em-pthread"));n.mountExternalData=(h,E)=>{h.startsWith("./")&&(h=h.substring(2)),(n.Eb||(n.Eb=new Map)).set(h,E)},n.unmountExternalData=()=>{delete n.Eb};var p=globalThis.SharedArrayBuffer??new WebAssembly.Memory({initial:0,maximum:0,pc:!0}).buffer.constructor;let d=h=>async(...E)=>{var I;try{if(n.Fb)throw Error("Session already started");let R=n.Fb={dc:E[0],errors:[]},H=await h(...E);if(n.Fb!==R)throw Error("Session mismatch");(I=n.Jb)==null||I.flush();let pe=R.errors;if(0<pe.length){let Ie=await Promise.all(pe);if(Ie=Ie.filter(Re=>Re),0<Ie.length)throw Error(Ie.join(`
`))}return H}finally{n.Fb=null}};n.jsepInit=(h,E)=>{if(h==="webgpu"){[n.Jb,n.Ub,n.Yb,n.Kb,n.Xb,n.jb,n.Zb,n.ac,n.Vb,n.Wb,n.$b]=E;let I=n.Jb;n.jsepRegisterBuffer=(R,H,pe,Ie)=>I.registerBuffer(R,H,pe,Ie),n.jsepGetBuffer=R=>I.getBuffer(R),n.jsepCreateDownloader=(R,H,pe)=>I.createDownloader(R,H,pe),n.jsepOnCreateSession=R=>{I.onCreateSession(R)},n.jsepOnReleaseSession=R=>{I.onReleaseSession(R)},n.jsepOnRunStart=R=>I.onRunStart(R),n.bc=(R,H)=>{I.upload(R,H)}}else if(h==="webnn"){let I=E[0];[n.nc,n.Nb,n.webnnEnsureTensor,n.Ob,n.webnnDownloadTensor]=E.slice(1),n.webnnReleaseTensorId=n.Nb,n.webnnUploadTensor=n.Ob,n.webnnOnRunStart=R=>I.onRunStart(R),n.webnnOnRunEnd=I.onRunEnd.bind(I),n.webnnRegisterMLContext=(R,H)=>{I.registerMLContext(R,H)},n.webnnOnReleaseSession=R=>{I.onReleaseSession(R)},n.webnnCreateMLTensorDownloader=(R,H)=>I.createMLTensorDownloader(R,H),n.webnnRegisterMLTensor=(R,H,pe,Ie)=>I.registerMLTensor(R,H,pe,Ie),n.webnnCreateMLContext=R=>I.createMLContext(R),n.webnnRegisterMLConstant=(R,H,pe,Ie,Re,Je)=>I.registerMLConstant(R,H,pe,Ie,Re,n.Eb,Je),n.webnnRegisterGraphInput=I.registerGraphInput.bind(I),n.webnnIsGraphInput=I.isGraphInput.bind(I),n.webnnCreateTemporaryTensor=I.createTemporaryTensor.bind(I),n.webnnIsInt64Supported=I.isInt64Supported.bind(I)}};let u=()=>{let h=(E,I,R)=>(...H)=>{let pe=wr,Ie=I==null?void 0:I();H=E(...H);let Re=I==null?void 0:I();return Ie!==Re&&(E=Re,R(Ie),I=R=null),wr!=pe?new Promise((Je,ot)=>{Zr={resolve:Je,reject:ot}}):H};(()=>{for(let E of["_OrtAppendExecutionProvider","_OrtCreateSession","_OrtRun","_OrtRunWithBinding","_OrtBindInput"])n[E]=h(n[E],()=>n[E],I=>n[E]=I)})(),d!==void 0&&(n._OrtRun=d(n._OrtRun),n._OrtRunWithBinding=d(n._OrtRunWithBinding)),u=void 0};n.asyncInit=()=>{u==null||u()};var _,f,b=Object.assign({},n),A=(h,E)=>{throw E},g="";(i||l)&&(l?g=self.location.href:typeof document<"u"&&document.currentScript&&(g=document.currentScript.src),Ri&&(g=Ri),g=g.startsWith("blob:")?"":g.slice(0,g.replace(/[?#].*/,"").lastIndexOf("/")+1),l&&(f=h=>{var E=new XMLHttpRequest;return E.open("GET",h,!1),E.responseType="arraybuffer",E.send(null),new Uint8Array(E.response)}),_=async h=>{if(B(h))return new Promise((I,R)=>{var H=new XMLHttpRequest;H.open("GET",h,!0),H.responseType="arraybuffer",H.onload=()=>{H.status==200||H.status==0&&H.response?I(H.response):R(H.status)},H.onerror=R,H.send(null)});var E=await fetch(h,{credentials:"same-origin"});if(E.ok)return E.arrayBuffer();throw Error(E.status+" : "+E.url)});var y=console.log.bind(console),C=console.error.bind(console),x=y,M=C;Object.assign(n,b),b=null;var T,v,P,F,D,K,U,j,ne,q,te,Z,ae,he=n.wasmBinary,Q=!1,B=h=>h.startsWith("file://");function O(){return T.buffer!=F.buffer&&z(),F}function W(){return T.buffer!=F.buffer&&z(),D}function N(){return T.buffer!=F.buffer&&z(),K}function J(){return T.buffer!=F.buffer&&z(),U}function ie(){return T.buffer!=F.buffer&&z(),j}function me(){return T.buffer!=F.buffer&&z(),ne}function Ae(){return T.buffer!=F.buffer&&z(),q}function Ve(){return T.buffer!=F.buffer&&z(),ae}if(c){let h=function(E){try{var I=E.data,R=I.Bb;if(R==="load"){let H=[];self.onmessage=pe=>H.push(pe),self.startWorker=()=>{postMessage({Bb:"loaded"});for(let pe of H)h(pe);self.onmessage=h};for(let pe of I.Rb)n[pe]&&!n[pe].proxy||(n[pe]=(...Ie)=>{postMessage({Bb:"callHandler",Qb:pe,args:Ie})},pe=="print"&&(x=n[pe]),pe=="printErr"&&(M=n[pe]));T=I.kc,z(),$e(I.lc)}else if(R==="run"){rn(I.Ab),Fn(I.Ab,0,0,1,0,0),Ur(),fe(I.Ab),X||(Uo(),X=!0);try{sn(I.fc,I.Hb)}catch(H){if(H!="unwind")throw H}}else I.target!=="setimmediate"&&(R==="checkMailbox"?X&&Se():R&&(M(`worker: received unknown command ${R}`),M(I)))}catch(H){throw Wo(),H}};var $e,X=!1;M=function(...E){E=E.join(" "),console.error(E)},self.alert=function(...E){postMessage({Bb:"alert",text:E.join(" "),ic:ln()})},self.onunhandledrejection=E=>{throw E.reason||E},self.onmessage=h}function z(){var h=T.buffer;n.HEAP8=F=new Int8Array(h),n.HEAP16=K=new Int16Array(h),n.HEAPU8=D=new Uint8Array(h),n.HEAPU16=U=new Uint16Array(h),n.HEAP32=j=new Int32Array(h),n.HEAPU32=ne=new Uint32Array(h),n.HEAPF32=q=new Float32Array(h),n.HEAPF64=ae=new Float64Array(h),n.HEAP64=te=new BigInt64Array(h),n.HEAPU64=Z=new BigUint64Array(h)}function _e(){c?startWorker(n):ut.Ca()}c||(T=new WebAssembly.Memory({initial:256,maximum:65536,shared:!0}),z());var Ee,Me=0,Ce=null;function ye(){if(--Me==0&&Ce){var h=Ce;Ce=null,h()}}function de(h){throw M(h="Aborted("+h+")"),Q=!0,h=new WebAssembly.RuntimeError(h+". Build with -sASSERTIONS for more info."),o(h),h}function we(){return{a:{L:Le,Aa:ke,b:ds,$:Os,A:$,pa:ee,X:oe,Z:xe,qa:De,na:nt,ga:wt,ma:pt,J:xt,Y:tt,V:It,oa:qt,W:Wr,va:kr,E:zs,Q:ms,O:Gr,D:je,u:rt,r:Qt,P:Hs,z:k,R:G,ja:se,T:Ge,aa:Qe,M:et,F:bt,ia:fe,sa:Rt,t:Gt,Ba:_r,w:_s,o:pr,l:Qr,c:Cs,n:Rs,j:$a,v:Aa,p:on,f:ka,s:Ia,m:Fa,e:Oa,k:Da,i:an,g:La,d:za,da:Ra,ea:Ba,fa:Na,ba:Xn,ca:Fo,N:Jn,xa:$u,ua:Oo,h:Ua,C:Wa,G:Ga,ta:Va,x:Ha,ra:Ka,U:qa,q:ja,y:Qa,K:Xa,S:Do,za:Ja,ya:Lo,ka:zo,la:Za,_:dt,B:Ro,I:eo,ha:Bo,H:No,a:T,wa:He}}}var ce={829644:(h,E,I,R,H)=>{if(n===void 0||!n.Eb)return 1;if((h=Kt(Number(h>>>0))).startsWith("./")&&(h=h.substring(2)),!(h=n.Eb.get(h)))return 2;if(E=Number(E>>>0),I=Number(I>>>0),R=Number(R>>>0),E+I>h.byteLength)return 3;try{let pe=h.subarray(E,E+I);switch(H){case 0:W().set(pe,R>>>0);break;case 1:n.mc?n.mc(R,pe):n.bc(R,pe);break;default:return 4}return 0}catch{return 4}},830468:(h,E,I)=>{n.Ob(h,W().subarray(E>>>0,E+I>>>0))},830532:()=>n.nc(),830574:h=>{n.Nb(h)},830611:()=>{n.Vb()},830642:()=>{n.Wb()},830671:()=>{n.$b()},830696:h=>n.Ub(h),830729:h=>n.Yb(h),830761:(h,E,I)=>{n.Kb(Number(h),Number(E),Number(I),!0)},830824:(h,E,I)=>{n.Kb(Number(h),Number(E),Number(I))},830881:()=>typeof wasmOffsetConverter<"u",830938:h=>{n.jb("Abs",h,void 0)},830989:h=>{n.jb("Neg",h,void 0)},831040:h=>{n.jb("Floor",h,void 0)},831093:h=>{n.jb("Ceil",h,void 0)},831145:h=>{n.jb("Reciprocal",h,void 0)},831203:h=>{n.jb("Sqrt",h,void 0)},831255:h=>{n.jb("Exp",h,void 0)},831306:h=>{n.jb("Erf",h,void 0)},831357:h=>{n.jb("Sigmoid",h,void 0)},831412:(h,E,I)=>{n.jb("HardSigmoid",h,{alpha:E,beta:I})},831491:h=>{n.jb("Log",h,void 0)},831542:h=>{n.jb("Sin",h,void 0)},831593:h=>{n.jb("Cos",h,void 0)},831644:h=>{n.jb("Tan",h,void 0)},831695:h=>{n.jb("Asin",h,void 0)},831747:h=>{n.jb("Acos",h,void 0)},831799:h=>{n.jb("Atan",h,void 0)},831851:h=>{n.jb("Sinh",h,void 0)},831903:h=>{n.jb("Cosh",h,void 0)},831955:h=>{n.jb("Asinh",h,void 0)},832008:h=>{n.jb("Acosh",h,void 0)},832061:h=>{n.jb("Atanh",h,void 0)},832114:h=>{n.jb("Tanh",h,void 0)},832166:h=>{n.jb("Not",h,void 0)},832217:(h,E,I)=>{n.jb("Clip",h,{min:E,max:I})},832286:h=>{n.jb("Clip",h,void 0)},832338:(h,E)=>{n.jb("Elu",h,{alpha:E})},832396:h=>{n.jb("Gelu",h,void 0)},832448:h=>{n.jb("Relu",h,void 0)},832500:(h,E)=>{n.jb("LeakyRelu",h,{alpha:E})},832564:(h,E)=>{n.jb("ThresholdedRelu",h,{alpha:E})},832634:(h,E)=>{n.jb("Cast",h,{to:E})},832692:h=>{n.jb("Add",h,void 0)},832743:h=>{n.jb("Sub",h,void 0)},832794:h=>{n.jb("Mul",h,void 0)},832845:h=>{n.jb("Div",h,void 0)},832896:h=>{n.jb("Pow",h,void 0)},832947:h=>{n.jb("Equal",h,void 0)},833e3:h=>{n.jb("Greater",h,void 0)},833055:h=>{n.jb("GreaterOrEqual",h,void 0)},833117:h=>{n.jb("Less",h,void 0)},833169:h=>{n.jb("LessOrEqual",h,void 0)},833228:(h,E,I,R,H)=>{n.jb("ReduceMean",h,{keepDims:!!E,noopWithEmptyAxes:!!I,axes:R?Array.from(ie().subarray(Number(R)>>>0,Number(H)>>>0)):[]})},833403:(h,E,I,R,H)=>{n.jb("ReduceMax",h,{keepDims:!!E,noopWithEmptyAxes:!!I,axes:R?Array.from(ie().subarray(Number(R)>>>0,Number(H)>>>0)):[]})},833577:(h,E,I,R,H)=>{n.jb("ReduceMin",h,{keepDims:!!E,noopWithEmptyAxes:!!I,axes:R?Array.from(ie().subarray(Number(R)>>>0,Number(H)>>>0)):[]})},833751:(h,E,I,R,H)=>{n.jb("ReduceProd",h,{keepDims:!!E,noopWithEmptyAxes:!!I,axes:R?Array.from(ie().subarray(Number(R)>>>0,Number(H)>>>0)):[]})},833926:(h,E,I,R,H)=>{n.jb("ReduceSum",h,{keepDims:!!E,noopWithEmptyAxes:!!I,axes:R?Array.from(ie().subarray(Number(R)>>>0,Number(H)>>>0)):[]})},834100:(h,E,I,R,H)=>{n.jb("ReduceL1",h,{keepDims:!!E,noopWithEmptyAxes:!!I,axes:R?Array.from(ie().subarray(Number(R)>>>0,Number(H)>>>0)):[]})},834273:(h,E,I,R,H)=>{n.jb("ReduceL2",h,{keepDims:!!E,noopWithEmptyAxes:!!I,axes:R?Array.from(ie().subarray(Number(R)>>>0,Number(H)>>>0)):[]})},834446:(h,E,I,R,H)=>{n.jb("ReduceLogSum",h,{keepDims:!!E,noopWithEmptyAxes:!!I,axes:R?Array.from(ie().subarray(Number(R)>>>0,Number(H)>>>0)):[]})},834623:(h,E,I,R,H)=>{n.jb("ReduceSumSquare",h,{keepDims:!!E,noopWithEmptyAxes:!!I,axes:R?Array.from(ie().subarray(Number(R)>>>0,Number(H)>>>0)):[]})},834803:(h,E,I,R,H)=>{n.jb("ReduceLogSumExp",h,{keepDims:!!E,noopWithEmptyAxes:!!I,axes:R?Array.from(ie().subarray(Number(R)>>>0,Number(H)>>>0)):[]})},834983:h=>{n.jb("Where",h,void 0)},835036:(h,E,I)=>{n.jb("Transpose",h,{perm:E?Array.from(ie().subarray(Number(E)>>>0,Number(I)>>>0)):[]})},835160:(h,E,I,R)=>{n.jb("DepthToSpace",h,{blocksize:E,mode:Kt(I),format:R?"NHWC":"NCHW"})},835293:(h,E,I,R)=>{n.jb("DepthToSpace",h,{blocksize:E,mode:Kt(I),format:R?"NHWC":"NCHW"})},835426:(h,E,I,R,H,pe,Ie,Re,Je,ot,Tt,zt,Ut,Tr,Qs)=>{n.jb("ConvTranspose",h,{format:Je?"NHWC":"NCHW",autoPad:E,dilations:[I],group:R,kernelShape:[H],pads:[pe,Ie],strides:[Re],wIsConst:()=>!!O()[ot>>>0],outputPadding:Tt?Array.from(ie().subarray(Number(Tt)>>>0,Number(zt)>>>0)):[],outputShape:Ut?Array.from(ie().subarray(Number(Ut)>>>0,Number(Tr)>>>0)):[],activation:Kt(Qs)})},835859:(h,E,I,R,H,pe,Ie,Re,Je,ot,Tt,zt,Ut,Tr)=>{n.jb("ConvTranspose",h,{format:Re?"NHWC":"NCHW",autoPad:E,dilations:Array.from(ie().subarray(Number(I)>>>0,2+(Number(I)>>>0)>>>0)),group:R,kernelShape:Array.from(ie().subarray(Number(H)>>>0,2+(Number(H)>>>0)>>>0)),pads:Array.from(ie().subarray(Number(pe)>>>0,4+(Number(pe)>>>0)>>>0)),strides:Array.from(ie().subarray(Number(Ie)>>>0,2+(Number(Ie)>>>0)>>>0)),wIsConst:()=>!!O()[Je>>>0],outputPadding:ot?Array.from(ie().subarray(Number(ot)>>>0,Number(Tt)>>>0)):[],outputShape:zt?Array.from(ie().subarray(Number(zt)>>>0,Number(Ut)>>>0)):[],activation:Kt(Tr)})},836520:(h,E,I,R,H,pe,Ie,Re,Je,ot,Tt,zt,Ut,Tr,Qs)=>{n.jb("ConvTranspose",h,{format:Je?"NHWC":"NCHW",autoPad:E,dilations:[I],group:R,kernelShape:[H],pads:[pe,Ie],strides:[Re],wIsConst:()=>!!O()[ot>>>0],outputPadding:Tt?Array.from(ie().subarray(Number(Tt)>>>0,Number(zt)>>>0)):[],outputShape:Ut?Array.from(ie().subarray(Number(Ut)>>>0,Number(Tr)>>>0)):[],activation:Kt(Qs)})},836953:(h,E,I,R,H,pe,Ie,Re,Je,ot,Tt,zt,Ut,Tr)=>{n.jb("ConvTranspose",h,{format:Re?"NHWC":"NCHW",autoPad:E,dilations:Array.from(ie().subarray(Number(I)>>>0,2+(Number(I)>>>0)>>>0)),group:R,kernelShape:Array.from(ie().subarray(Number(H)>>>0,2+(Number(H)>>>0)>>>0)),pads:Array.from(ie().subarray(Number(pe)>>>0,4+(Number(pe)>>>0)>>>0)),strides:Array.from(ie().subarray(Number(Ie)>>>0,2+(Number(Ie)>>>0)>>>0)),wIsConst:()=>!!O()[Je>>>0],outputPadding:ot?Array.from(ie().subarray(Number(ot)>>>0,Number(Tt)>>>0)):[],outputShape:zt?Array.from(ie().subarray(Number(zt)>>>0,Number(Ut)>>>0)):[],activation:Kt(Tr)})},837614:(h,E)=>{n.jb("GlobalAveragePool",h,{format:E?"NHWC":"NCHW"})},837705:(h,E,I,R,H,pe,Ie,Re,Je,ot,Tt,zt,Ut,Tr)=>{n.jb("AveragePool",h,{format:Tr?"NHWC":"NCHW",auto_pad:E,ceil_mode:I,count_include_pad:R,storage_order:H,dilations:pe?Array.from(ie().subarray(Number(pe)>>>0,Number(Ie)>>>0)):[],kernel_shape:Re?Array.from(ie().subarray(Number(Re)>>>0,Number(Je)>>>0)):[],pads:ot?Array.from(ie().subarray(Number(ot)>>>0,Number(Tt)>>>0)):[],strides:zt?Array.from(ie().subarray(Number(zt)>>>0,Number(Ut)>>>0)):[]})},838184:(h,E)=>{n.jb("GlobalAveragePool",h,{format:E?"NHWC":"NCHW"})},838275:(h,E,I,R,H,pe,Ie,Re,Je,ot,Tt,zt,Ut,Tr)=>{n.jb("AveragePool",h,{format:Tr?"NHWC":"NCHW",auto_pad:E,ceil_mode:I,count_include_pad:R,storage_order:H,dilations:pe?Array.from(ie().subarray(Number(pe)>>>0,Number(Ie)>>>0)):[],kernel_shape:Re?Array.from(ie().subarray(Number(Re)>>>0,Number(Je)>>>0)):[],pads:ot?Array.from(ie().subarray(Number(ot)>>>0,Number(Tt)>>>0)):[],strides:zt?Array.from(ie().subarray(Number(zt)>>>0,Number(Ut)>>>0)):[]})},838754:(h,E)=>{n.jb("GlobalMaxPool",h,{format:E?"NHWC":"NCHW"})},838841:(h,E,I,R,H,pe,Ie,Re,Je,ot,Tt,zt,Ut,Tr)=>{n.jb("MaxPool",h,{format:Tr?"NHWC":"NCHW",auto_pad:E,ceil_mode:I,count_include_pad:R,storage_order:H,dilations:pe?Array.from(ie().subarray(Number(pe)>>>0,Number(Ie)>>>0)):[],kernel_shape:Re?Array.from(ie().subarray(Number(Re)>>>0,Number(Je)>>>0)):[],pads:ot?Array.from(ie().subarray(Number(ot)>>>0,Number(Tt)>>>0)):[],strides:zt?Array.from(ie().subarray(Number(zt)>>>0,Number(Ut)>>>0)):[]})},839316:(h,E)=>{n.jb("GlobalMaxPool",h,{format:E?"NHWC":"NCHW"})},839403:(h,E,I,R,H,pe,Ie,Re,Je,ot,Tt,zt,Ut,Tr)=>{n.jb("MaxPool",h,{format:Tr?"NHWC":"NCHW",auto_pad:E,ceil_mode:I,count_include_pad:R,storage_order:H,dilations:pe?Array.from(ie().subarray(Number(pe)>>>0,Number(Ie)>>>0)):[],kernel_shape:Re?Array.from(ie().subarray(Number(Re)>>>0,Number(Je)>>>0)):[],pads:ot?Array.from(ie().subarray(Number(ot)>>>0,Number(Tt)>>>0)):[],strides:zt?Array.from(ie().subarray(Number(zt)>>>0,Number(Ut)>>>0)):[]})},839878:(h,E,I,R,H)=>{n.jb("Gemm",h,{alpha:E,beta:I,transA:R,transB:H})},839982:h=>{n.jb("MatMul",h,void 0)},840036:(h,E,I,R)=>{n.jb("ArgMax",h,{keepDims:!!E,selectLastIndex:!!I,axis:R})},840144:(h,E,I,R)=>{n.jb("ArgMin",h,{keepDims:!!E,selectLastIndex:!!I,axis:R})},840252:(h,E)=>{n.jb("Softmax",h,{axis:E})},840315:(h,E)=>{n.jb("Concat",h,{axis:E})},840375:(h,E,I,R,H)=>{n.jb("Split",h,{axis:E,numOutputs:I,splitSizes:R?Array.from(ie().subarray(Number(R)>>>0,Number(H)>>>0)):[]})},840531:h=>{n.jb("Expand",h,void 0)},840585:(h,E)=>{n.jb("Gather",h,{axis:Number(E)})},840656:(h,E)=>{n.jb("GatherElements",h,{axis:Number(E)})},840735:(h,E)=>{n.jb("GatherND",h,{batch_dims:Number(E)})},840814:(h,E,I,R,H,pe,Ie,Re,Je,ot,Tt)=>{n.jb("Resize",h,{antialias:E,axes:I?Array.from(ie().subarray(Number(I)>>>0,Number(R)>>>0)):[],coordinateTransformMode:Kt(H),cubicCoeffA:pe,excludeOutside:Ie,extrapolationValue:Re,keepAspectRatioPolicy:Kt(Je),mode:Kt(ot),nearestMode:Kt(Tt)})},841176:(h,E,I,R,H,pe,Ie)=>{n.jb("Slice",h,{starts:E?Array.from(ie().subarray(Number(E)>>>0,Number(I)>>>0)):[],ends:R?Array.from(ie().subarray(Number(R)>>>0,Number(H)>>>0)):[],axes:pe?Array.from(ie().subarray(Number(pe)>>>0,Number(Ie)>>>0)):[]})},841440:h=>{n.jb("Tile",h,void 0)},841492:(h,E,I)=>{n.jb("InstanceNormalization",h,{epsilon:E,format:I?"NHWC":"NCHW"})},841606:(h,E,I)=>{n.jb("InstanceNormalization",h,{epsilon:E,format:I?"NHWC":"NCHW"})},841720:h=>{n.jb("Range",h,void 0)},841773:(h,E)=>{n.jb("Einsum",h,{equation:Kt(E)})},841854:(h,E,I,R,H)=>{n.jb("Pad",h,{mode:E,value:I,pads:R?Array.from(ie().subarray(Number(R)>>>0,Number(H)>>>0)):[]})},841997:(h,E,I,R,H,pe)=>{n.jb("BatchNormalization",h,{epsilon:E,momentum:I,spatial:!!H,trainingMode:!!R,format:pe?"NHWC":"NCHW"})},842166:(h,E,I,R,H,pe)=>{n.jb("BatchNormalization",h,{epsilon:E,momentum:I,spatial:!!H,trainingMode:!!R,format:pe?"NHWC":"NCHW"})},842335:(h,E,I)=>{n.jb("CumSum",h,{exclusive:Number(E),reverse:Number(I)})},842432:(h,E,I)=>{n.jb("DequantizeLinear",h,{axis:E,blockSize:I})},842522:(h,E,I,R,H)=>{n.jb("GridSample",h,{align_corners:E,mode:Kt(I),padding_mode:Kt(R),format:H?"NHWC":"NCHW"})},842692:(h,E,I,R,H)=>{n.jb("GridSample",h,{align_corners:E,mode:Kt(I),padding_mode:Kt(R),format:H?"NHWC":"NCHW"})},842862:(h,E)=>{n.jb("ScatterND",h,{reduction:Kt(E)})},842947:(h,E,I,R,H,pe,Ie,Re,Je)=>{n.jb("Attention",h,{numHeads:E,isUnidirectional:I,maskFilterValue:R,scale:H,doRotary:pe,qkvHiddenSizes:Ie?Array.from(ie().subarray(Number(Re)>>>0,Number(Re)+Ie>>>0)):[],pastPresentShareBuffer:!!Je})},843219:h=>{n.jb("BiasAdd",h,void 0)},843274:h=>{n.jb("BiasSplitGelu",h,void 0)},843335:h=>{n.jb("FastGelu",h,void 0)},843391:(h,E,I,R,H,pe,Ie,Re,Je,ot,Tt,zt,Ut,Tr,Qs,Yo)=>{n.jb("Conv",h,{format:zt?"NHWC":"NCHW",auto_pad:E,dilations:I?Array.from(ie().subarray(Number(I)>>>0,Number(R)>>>0)):[],group:H,kernel_shape:pe?Array.from(ie().subarray(Number(pe)>>>0,Number(Ie)>>>0)):[],pads:Re?Array.from(ie().subarray(Number(Re)>>>0,Number(Je)>>>0)):[],strides:ot?Array.from(ie().subarray(Number(ot)>>>0,Number(Tt)>>>0)):[],w_is_const:()=>!!O()[Number(Ut)>>>0],activation:Kt(Tr),activation_params:Qs?Array.from(Ae().subarray(Number(Qs)>>>0,Number(Yo)>>>0)):[]})},843975:h=>{n.jb("Gelu",h,void 0)},844027:(h,E,I,R,H,pe,Ie,Re,Je)=>{n.jb("GroupQueryAttention",h,{numHeads:E,kvNumHeads:I,scale:R,softcap:H,doRotary:pe,rotaryInterleaved:Ie,smoothSoftmax:Re,localWindowSize:Je})},844244:(h,E,I,R)=>{n.jb("LayerNormalization",h,{axis:E,epsilon:I,simplified:!!R})},844355:(h,E,I,R)=>{n.jb("LayerNormalization",h,{axis:E,epsilon:I,simplified:!!R})},844466:(h,E,I,R,H,pe)=>{n.jb("MatMulNBits",h,{k:E,n:I,accuracyLevel:R,bits:H,blockSize:pe})},844593:(h,E,I,R,H,pe)=>{n.jb("MultiHeadAttention",h,{numHeads:E,isUnidirectional:I,maskFilterValue:R,scale:H,doRotary:pe})},844752:(h,E)=>{n.jb("QuickGelu",h,{alpha:E})},844816:(h,E,I,R,H)=>{n.jb("RotaryEmbedding",h,{interleaved:!!E,numHeads:I,rotaryEmbeddingDim:R,scale:H})},844955:(h,E,I)=>{n.jb("SkipLayerNormalization",h,{epsilon:E,simplified:!!I})},845057:(h,E,I)=>{n.jb("SkipLayerNormalization",h,{epsilon:E,simplified:!!I})},845159:(h,E,I,R)=>{n.jb("GatherBlockQuantized",h,{gatherAxis:E,quantizeAxis:I,blockSize:R})},845280:h=>{n.Zb(h)},845314:(h,E)=>n.ac(Number(h),Number(E),n.Fb.dc,n.Fb.errors)};function ke(h,E,I){return Or(async()=>{await n.Xb(Number(h),Number(E),Number(I))})}function Le(){return typeof wasmOffsetConverter<"u"}class Te{constructor(E){re(this,"name","ExitStatus");this.message=`Program terminated with exit(${E})`,this.status=E}}var We=h=>{h.terminate(),h.onmessage=()=>{}},qe=[],st=h=>{kt.length==0&&(Ar(),sr(kt[0]));var E=kt.pop();if(!E)return 6;ht.push(E),$r[h.Ab]=E,E.Ab=h.Ab;var I={Bb:"run",fc:h.ec,Hb:h.Hb,Ab:h.Ab};return E.postMessage(I,h.Mb),0},Ze=0,ze=(h,E,...I)=>{for(var R=2*I.length,H=no(),pe=Dn(8*R),Ie=pe>>>3,Re=0;Re<I.length;Re++){var Je=I[Re];typeof Je=="bigint"?(te[Ie+2*Re]=1n,te[Ie+2*Re+1]=Je):(te[Ie+2*Re]=0n,Ve()[Ie+2*Re+1>>>0]=Je)}return h=Go(h,0,R,pe,E),On(H),h};function He(h){if(c)return ze(0,1,h);if(P=h,!(0<Ze)){for(var E of ht)We(E);for(E of kt)We(E);kt=[],ht=[],$r={},Q=!0}A(0,new Te(h))}function gt(h){if(c)return ze(1,0,h);dt(h)}var dt=h=>{if(P=h,c)throw gt(h),"unwind";He(h)},kt=[],ht=[],yr=[],$r={},Vr=h=>{var E=h.Ab;delete $r[E],kt.push(h),ht.splice(ht.indexOf(h),1),h.Ab=0,to(E)};function Ur(){yr.forEach(h=>h())}var sr=h=>new Promise(E=>{h.onmessage=H=>{var pe=(H=H.data).Bb;if(H.Gb&&H.Gb!=ln()){var Ie=$r[H.Gb];Ie?Ie.postMessage(H,H.Mb):M(`Internal error! Worker sent a message "${pe}" to target pthread ${H.Gb}, but that thread no longer exists!`)}else pe==="checkMailbox"?Se():pe==="spawnThread"?st(H):pe==="cleanupThread"?Vr($r[H.hc]):pe==="loaded"?(h.loaded=!0,E(h)):pe==="alert"?alert(`Thread ${H.ic}: ${H.text}`):H.target==="setimmediate"?h.postMessage(H):pe==="callHandler"?n[H.Qb](...H.args):pe&&M(`worker sent an unknown command ${pe}`)},h.onerror=H=>{throw M(`worker sent an error! ${H.filename}:${H.lineno}: ${H.message}`),H};var I,R=[];for(I of[])n.propertyIsEnumerable(I)&&R.push(I);h.postMessage({Bb:"load",Rb:R,kc:T,lc:v})});function Ar(){var h=new Worker((()=>{let E=URL;return self.location.href>"file:"&&self.location.href<"file;"?new E("ort.bundle.min.mjs",self.location.href):new URL(self.location.href)})(),{type:"module",workerData:"em-pthread",name:"em-pthread"});kt.push(h)}var rn=h=>{z();var E=me()[h+52>>>2>>>0];h=me()[h+56>>>2>>>0],Ko(E,E-h),On(E)},sn=(h,E)=>{Ze=0,h=qo(h,E),0<Ze?P=h:ro(h)};class nn{constructor(E){this.Ib=E-24}}function ds(h,E,I){var R=new nn(h>>>=0);throw E>>>=0,I>>>=0,me()[R.Ib+16>>>2>>>0]=0,me()[R.Ib+4>>>2>>>0]=E,me()[R.Ib+8>>>2>>>0]=I,h}function ft(h,E,I,R){return c?ze(2,1,h,E,I,R):Os(h,E,I,R)}function Os(h,E,I,R){if(h>>>=0,I>>>=0,R>>>=0,p===void 0)return 6;var H=[];return c&&H.length===0?ft(h,E>>>=0,I,R):(h={ec:I,Ab:h,Hb:R,Mb:H},c?(h.Bb="spawnThread",postMessage(h,H),0):st(h))}var Ds=typeof TextDecoder<"u"?new TextDecoder:void 0,St=(h,E=0,I=NaN)=>{var R=(E>>>=0)+I;for(I=E;h[I]&&!(I>=R);)++I;if(16<I-E&&h.buffer&&Ds)return Ds.decode(h.buffer instanceof ArrayBuffer?h.subarray(E,I):h.slice(E,I));for(R="";E<I;){var H=h[E++];if(128&H){var pe=63&h[E++];if((224&H)==192)R+=String.fromCharCode((31&H)<<6|pe);else{var Ie=63&h[E++];65536>(H=(240&H)==224?(15&H)<<12|pe<<6|Ie:(7&H)<<18|pe<<12|Ie<<6|63&h[E++])?R+=String.fromCharCode(H):(H-=65536,R+=String.fromCharCode(55296|H>>10,56320|1023&H))}}else R+=String.fromCharCode(H)}return R},Kt=(h,E)=>(h>>>=0)?St(W(),h,E):"";function $(h,E,I){return c?ze(3,1,h,E,I):0}function ee(h,E){if(c)return ze(4,1,h,E)}var V=h=>{for(var E=0,I=0;I<h.length;++I){var R=h.charCodeAt(I);127>=R?E++:2047>=R?E+=2:55296<=R&&57343>=R?(E+=4,++I):E+=3}return E},Y=(h,E,I)=>{var R=W();if(E>>>=0,0<I){var H=E;I=E+I-1;for(var pe=0;pe<h.length;++pe){var Ie=h.charCodeAt(pe);if(55296<=Ie&&57343>=Ie&&(Ie=65536+((1023&Ie)<<10)|1023&h.charCodeAt(++pe)),127>=Ie){if(E>=I)break;R[E++>>>0]=Ie}else{if(2047>=Ie){if(E+1>=I)break;R[E++>>>0]=192|Ie>>6}else{if(65535>=Ie){if(E+2>=I)break;R[E++>>>0]=224|Ie>>12}else{if(E+3>=I)break;R[E++>>>0]=240|Ie>>18,R[E++>>>0]=128|Ie>>12&63}R[E++>>>0]=128|Ie>>6&63}R[E++>>>0]=128|63&Ie}}R[E>>>0]=0,h=E-H}else h=0;return h};function oe(h,E){if(c)return ze(5,1,h,E)}function xe(h,E,I){if(c)return ze(6,1,h,E,I)}function De(h,E,I){return c?ze(7,1,h,E,I):0}function nt(h,E){if(c)return ze(8,1,h,E)}function wt(h,E,I){if(c)return ze(9,1,h,E,I)}function pt(h,E,I,R){if(c)return ze(10,1,h,E,I,R)}function xt(h,E,I,R){if(c)return ze(11,1,h,E,I,R)}function tt(h,E,I,R){if(c)return ze(12,1,h,E,I,R)}function It(h){if(c)return ze(13,1,h)}function qt(h,E){if(c)return ze(14,1,h,E)}function Wr(h,E,I){if(c)return ze(15,1,h,E,I)}var qr,nr,kr=()=>de(""),cr=h=>{for(var E="";W()[h>>>0];)E+=qr[W()[h++>>>0]];return E},ps={},hs={};function Ir(h,E,I={}){return function(R,H,pe={}){var Ie=H.name;if(!R)throw new nr(`type "${Ie}" must have a positive integer typeid pointer`);if(hs.hasOwnProperty(R)){if(pe.Sb)return;throw new nr(`Cannot register type '${Ie}' twice`)}hs[R]=H,ps.hasOwnProperty(R)&&(H=ps[R],delete ps[R],H.forEach(Re=>Re()))}(h,E,I)}var Ls=(h,E,I)=>{switch(E){case 1:return I?R=>O()[R>>>0]:R=>W()[R>>>0];case 2:return I?R=>N()[R>>>1>>>0]:R=>J()[R>>>1>>>0];case 4:return I?R=>ie()[R>>>2>>>0]:R=>me()[R>>>2>>>0];case 8:return I?R=>te[R>>>3]:R=>Z[R>>>3];default:throw new TypeError(`invalid integer width (${E}): ${h}`)}};function zs(h,E,I){I>>>=0,Ir(h>>>=0,{name:E=cr(E>>>0),fromWireType:R=>R,toWireType:function(R,H){if(typeof H!="bigint"&&typeof H!="number")throw H=H===null?"null":(R=typeof H)=="object"||R==="array"||R==="function"?H.toString():""+H,new TypeError(`Cannot convert "${H}" to ${this.name}`);return typeof H=="number"&&(H=BigInt(H)),H},Cb:vr,readValueFromPointer:Ls(E,I,E.indexOf("u")==-1),Db:null})}var vr=8;function ms(h,E,I,R){Ir(h>>>=0,{name:E=cr(E>>>0),fromWireType:function(H){return!!H},toWireType:function(H,pe){return pe?I:R},Cb:vr,readValueFromPointer:function(H){return this.fromWireType(W()[H>>>0])},Db:null})}var Yr=[],Rr=[];function Cs(h){9<(h>>>=0)&&--Rr[h+1]==0&&(Rr[h]=void 0,Yr.push(h))}var mr=h=>{if(!h)throw new nr("Cannot use deleted val. handle = "+h);return Rr[h]},ar=h=>{switch(h){case void 0:return 2;case null:return 4;case!0:return 6;case!1:return 8;default:let E=Yr.pop()||Rr.length;return Rr[E]=h,Rr[E+1]=1,E}};function fs(h){return this.fromWireType(me()[h>>>2>>>0])}var Gs={name:"emscripten::val",fromWireType:h=>{var E=mr(h);return Cs(h),E},toWireType:(h,E)=>ar(E),Cb:vr,readValueFromPointer:fs,Db:null};function Gr(h){return Ir(h>>>0,Gs)}var Ne=(h,E)=>{switch(E){case 4:return function(I){return this.fromWireType(Ae()[I>>>2>>>0])};case 8:return function(I){return this.fromWireType(Ve()[I>>>3>>>0])};default:throw new TypeError(`invalid float width (${E}): ${h}`)}};function je(h,E,I){I>>>=0,Ir(h>>>=0,{name:E=cr(E>>>0),fromWireType:R=>R,toWireType:(R,H)=>H,Cb:vr,readValueFromPointer:Ne(E,I),Db:null})}function rt(h,E,I,R,H){if(h>>>=0,I>>>=0,E=cr(E>>>0),H===-1&&(H=4294967295),H=Re=>Re,R===0){var pe=32-8*I;H=Re=>Re<<pe>>>pe}var Ie=E.includes("unsigned")?function(Re,Je){return Je>>>0}:function(Re,Je){return Je};Ir(h,{name:E,fromWireType:H,toWireType:Ie,Cb:vr,readValueFromPointer:Ls(E,I,R!==0),Db:null})}function Qt(h,E,I){function R(pe){var Ie=me()[pe>>>2>>>0];return pe=me()[pe+4>>>2>>>0],new H(O().buffer,pe,Ie)}var H=[Int8Array,Uint8Array,Int16Array,Uint16Array,Int32Array,Uint32Array,Float32Array,Float64Array,BigInt64Array,BigUint64Array][E];Ir(h>>>=0,{name:I=cr(I>>>0),fromWireType:R,Cb:vr,readValueFromPointer:R},{Sb:!0})}function Hs(h,E){Ir(h>>>=0,{name:E=cr(E>>>0),fromWireType:function(I){for(var R,H=me()[I>>>2>>>0],pe=I+4,Ie=pe,Re=0;Re<=H;++Re){var Je=pe+Re;Re!=H&&W()[Je>>>0]!=0||(Ie=Kt(Ie,Je-Ie),R===void 0?R=Ie:(R+="\0",R+=Ie),Ie=Je+1)}return os(I),R},toWireType:function(I,R){R instanceof ArrayBuffer&&(R=new Uint8Array(R));var H=typeof R=="string";if(!(H||R instanceof Uint8Array||R instanceof Uint8ClampedArray||R instanceof Int8Array))throw new nr("Cannot pass non-string to std::string");var pe=H?V(R):R.length,Ie=In(4+pe+1),Re=Ie+4;if(me()[Ie>>>2>>>0]=pe,H)Y(R,Re,pe+1);else if(H)for(H=0;H<pe;++H){var Je=R.charCodeAt(H);if(255<Je)throw os(Ie),new nr("String has UTF-16 code units that do not fit in 8 bits");W()[Re+H>>>0]=Je}else for(H=0;H<pe;++H)W()[Re+H>>>0]=R[H];return I!==null&&I.push(os,Ie),Ie},Cb:vr,readValueFromPointer:fs,Db(I){os(I)}})}var Ss=typeof TextDecoder<"u"?new TextDecoder("utf-16le"):void 0,ss=(h,E)=>{for(var I=h>>1,R=I+E/2;!(I>=R)&&J()[I>>>0];)++I;if(32<(I<<=1)-h&&Ss)return Ss.decode(W().slice(h,I));for(I="",R=0;!(R>=E/2);++R){var H=N()[h+2*R>>>1>>>0];if(H==0)break;I+=String.fromCharCode(H)}return I},Tn=(h,E,I)=>{if(I??(I=2147483647),2>I)return 0;var R=E;I=(I-=2)<2*h.length?I/2:h.length;for(var H=0;H<I;++H){var pe=h.charCodeAt(H);N()[E>>>1>>>0]=pe,E+=2}return N()[E>>>1>>>0]=0,E-R},En=h=>2*h.length,Pn=(h,E)=>{for(var I=0,R="";!(I>=E/4);){var H=ie()[h+4*I>>>2>>>0];if(H==0)break;++I,65536<=H?(H-=65536,R+=String.fromCharCode(55296|H>>10,56320|1023&H)):R+=String.fromCharCode(H)}return R},Cn=(h,E,I)=>{if(E>>>=0,I??(I=2147483647),4>I)return 0;var R=E;I=R+I-4;for(var H=0;H<h.length;++H){var pe=h.charCodeAt(H);if(55296<=pe&&57343>=pe&&(pe=65536+((1023&pe)<<10)|1023&h.charCodeAt(++H)),ie()[E>>>2>>>0]=pe,(E+=4)+4>I)break}return ie()[E>>>2>>>0]=0,E-R},ge=h=>{for(var E=0,I=0;I<h.length;++I){var R=h.charCodeAt(I);55296<=R&&57343>=R&&++I,E+=4}return E};function k(h,E,I){if(h>>>=0,E>>>=0,I=cr(I>>>=0),E===2)var R=ss,H=Tn,pe=En,Ie=Re=>J()[Re>>>1>>>0];else E===4&&(R=Pn,H=Cn,pe=ge,Ie=Re=>me()[Re>>>2>>>0]);Ir(h,{name:I,fromWireType:Re=>{for(var Je,ot=me()[Re>>>2>>>0],Tt=Re+4,zt=0;zt<=ot;++zt){var Ut=Re+4+zt*E;zt!=ot&&Ie(Ut)!=0||(Tt=R(Tt,Ut-Tt),Je===void 0?Je=Tt:(Je+="\0",Je+=Tt),Tt=Ut+E)}return os(Re),Je},toWireType:(Re,Je)=>{if(typeof Je!="string")throw new nr(`Cannot pass non-string to C++ string type ${I}`);var ot=pe(Je),Tt=In(4+ot+E);return me()[Tt>>>2>>>0]=ot/E,H(Je,Tt+4,ot+E),Re!==null&&Re.push(os,Tt),Tt},Cb:vr,readValueFromPointer:fs,Db(Re){os(Re)}})}function G(h,E){Ir(h>>>=0,{Tb:!0,name:E=cr(E>>>0),Cb:0,fromWireType:()=>{},toWireType:()=>{}})}function se(h){Fn(h>>>0,!l,1,!i,131072,!1),Ur()}var ue=h=>{if(!Q)try{if(h(),!(0<Ze))try{c?ro(P):dt(P)}catch(E){E instanceof Te||E=="unwind"||A(0,E)}}catch(E){E instanceof Te||E=="unwind"||A(0,E)}};function fe(h){h>>>=0,typeof Atomics.jc=="function"&&(Atomics.jc(ie(),h>>>2,h).value.then(Se),h+=128,Atomics.store(ie(),h>>>2,1))}var Se=()=>{var h=ln();h&&(fe(h),ue(so))};function Ge(h,E){(h>>>=0)==E>>>0?setTimeout(Se):c?postMessage({Gb:h,Bb:"checkMailbox"}):(h=$r[h])&&h.postMessage({Bb:"checkMailbox"})}var Xe=[];function Qe(h,E,I,R,H){for(E>>>=0,R/=2,Xe.length=R,I=H>>>0>>>3,H=0;H<R;H++)Xe[H]=te[I+2*H]?te[I+2*H+1]:Ve()[I+2*H+1>>>0];return(E?ce[E]:ti[h])(...Xe)}var et=()=>{Ze=0};function bt(h){h>>>=0,c?postMessage({Bb:"cleanupThread",hc:h}):Vr($r[h])}function Rt(h){}var Lt=(h,E)=>{var I=hs[h];if(I===void 0)throw h=Vo(h),I=cr(h),os(h),new nr(`${E} has unknown type ${I}`);return I},Zt=(h,E,I)=>{var R=[];return h=h.toWireType(R,I),R.length&&(me()[E>>>2>>>0]=ar(R)),h};function Gt(h,E,I){return E>>>=0,I>>>=0,h=mr(h>>>0),E=Lt(E,"emval::as"),Zt(E,I,h)}function _r(h,E){return E>>>=0,h=mr(h>>>0),(E=Lt(E,"emval::as")).toWireType(null,h)}var gr=h=>{try{h()}catch(E){de(E)}},dr=0,wr=null,xr=0,ns=[],Yt={},Mr={},Fr=0,Zr=null,$s=[];function Or(h){return function(E){if(!Q){if(dr===0){var I=!1,R=!1;E((H=0)=>{if(!Q&&(xr=H,I=!0,R)){dr=2,gr(()=>Xo(wr)),typeof MainLoop<"u"&&MainLoop.Pb&&MainLoop.resume(),H=!1;try{var pe=function(){var Je=ie()[wr+8>>>2>>>0];return Je=ut[Mr[Je]],--Ze,Je()}()}catch(Je){pe=Je,H=!0}var Ie=!1;if(!wr){var Re=Zr;Re&&(Zr=null,(H?Re.reject:Re.resolve)(pe),Ie=!0)}if(H&&!Ie)throw pe}}),R=!0,I||(dr=1,wr=function(){var H=In(65548),pe=H+12;me()[H>>>2>>>0]=pe,me()[H+4>>>2>>>0]=pe+65536,pe=ns[0];var Ie=Yt[pe];return Ie===void 0&&(Ie=Fr++,Yt[pe]=Ie,Mr[Ie]=pe),pe=Ie,ie()[H+8>>>2>>>0]=pe,H}(),typeof MainLoop<"u"&&MainLoop.Pb&&MainLoop.pause(),gr(()=>oo(wr)))}else dr===2?(dr=0,gr(ao),os(wr),wr=null,$s.forEach(ue)):de(`invalid state: ${dr}`);return xr}}(E=>{h().then(E)})}function _s(h){return h>>>=0,Or(async()=>{var E=await mr(h);return ar(E)})}var ir=[];function pr(h,E,I,R){return I>>>=0,R>>>=0,(h=ir[h>>>0])(null,E=mr(E>>>0),I,R)}var fr={},er=h=>{var E=fr[h];return E===void 0?cr(h):E};function Qr(h,E,I,R,H){return I>>>=0,R>>>=0,H>>>=0,(h=ir[h>>>0])(E=mr(E>>>0),E[I=er(I)],R,H)}var Ks=()=>typeof globalThis=="object"?globalThis:Function("return this")();function Rs(h){return(h>>>=0)==0?ar(Ks()):(h=er(h),ar(Ks()[h]))}var Ca=h=>{var E=ir.length;return ir.push(h),E},Sa=(h,E)=>{for(var I=Array(h),R=0;R<h;++R)I[R]=Lt(me()[E+4*R>>>2>>>0],"parameter "+R);return I},Ao=(h,E)=>Object.defineProperty(E,"name",{value:h});function $a(h,E,I){var R=(E=Sa(h,E>>>0)).shift();h--;var H=`return function (obj, func, destructorsRef, args) {
`,pe=0,Ie=[];I===0&&Ie.push("obj");for(var Re=["retType"],Je=[R],ot=0;ot<h;++ot)Ie.push("arg"+ot),Re.push("argType"+ot),Je.push(E[ot]),H+=`  var arg${ot} = argType${ot}.readValueFromPointer(args${pe?"+"+pe:""});
`,pe+=E[ot].Cb;return H+=`  var rv = ${I===1?"new func":"func.call"}(${Ie.join(", ")});
`,R.Tb||(Re.push("emval_returnValue"),Je.push(Zt),H+=`  return emval_returnValue(retType, destructorsRef, rv);
`),Re.push(H+`};
`),h=function(Tt){var zt=Function;if(!(zt instanceof Function))throw new TypeError(`new_ called with constructor type ${typeof zt} which is not a function`);var Ut=Ao(zt.name||"unknownFunctionName",function(){});return Ut.prototype=zt.prototype,Ut=new Ut,(Tt=zt.apply(Ut,Tt))instanceof Object?Tt:Ut}(Re)(...Je),I=`methodCaller<(${E.map(Tt=>Tt.name).join(", ")}) => ${R.name}>`,Ca(Ao(I,h))}function Aa(h){return h=er(h>>>0),ar(n[h])}function on(h,E){return E>>>=0,h=mr(h>>>0),E=mr(E),ar(h[E])}function ka(h){9<(h>>>=0)&&(Rr[h+1]+=1)}function Ia(){return ar([])}function Fa(h){h=mr(h>>>0);for(var E=Array(h.length),I=0;I<h.length;I++)E[I]=h[I];return ar(E)}function Oa(h){return ar(er(h>>>0))}function Da(){return ar({})}function an(h){for(var E=mr(h>>>=0);E.length;){var I=E.pop();E.pop()(I)}Cs(h)}function La(h,E,I){E>>>=0,I>>>=0,h=mr(h>>>0),E=mr(E),I=mr(I),h[E]=I}function za(h,E){return E>>>=0,h=(h=Lt(h>>>0,"_emval_take_value")).readValueFromPointer(E),ar(h)}function Ra(h,E){h=-9007199254740992>h||9007199254740992<h?NaN:Number(h),E>>>=0,h=new Date(1e3*h),ie()[E>>>2>>>0]=h.getUTCSeconds(),ie()[E+4>>>2>>>0]=h.getUTCMinutes(),ie()[E+8>>>2>>>0]=h.getUTCHours(),ie()[E+12>>>2>>>0]=h.getUTCDate(),ie()[E+16>>>2>>>0]=h.getUTCMonth(),ie()[E+20>>>2>>>0]=h.getUTCFullYear()-1900,ie()[E+24>>>2>>>0]=h.getUTCDay(),h=(h.getTime()-Date.UTC(h.getUTCFullYear(),0,1,0,0,0,0))/864e5|0,ie()[E+28>>>2>>>0]=h}var ko=h=>h%4==0&&(h%100!=0||h%400==0),Io=[0,31,60,91,121,152,182,213,244,274,305,335],Qn=[0,31,59,90,120,151,181,212,243,273,304,334];function Ba(h,E){h=-9007199254740992>h||9007199254740992<h?NaN:Number(h),E>>>=0,h=new Date(1e3*h),ie()[E>>>2>>>0]=h.getSeconds(),ie()[E+4>>>2>>>0]=h.getMinutes(),ie()[E+8>>>2>>>0]=h.getHours(),ie()[E+12>>>2>>>0]=h.getDate(),ie()[E+16>>>2>>>0]=h.getMonth(),ie()[E+20>>>2>>>0]=h.getFullYear()-1900,ie()[E+24>>>2>>>0]=h.getDay();var I=(ko(h.getFullYear())?Io:Qn)[h.getMonth()]+h.getDate()-1|0;ie()[E+28>>>2>>>0]=I,ie()[E+36>>>2>>>0]=-60*h.getTimezoneOffset(),I=new Date(h.getFullYear(),6,1).getTimezoneOffset();var R=new Date(h.getFullYear(),0,1).getTimezoneOffset();h=0|(I!=R&&h.getTimezoneOffset()==Math.min(R,I)),ie()[E+32>>>2>>>0]=h}function Na(h){h>>>=0;var E=new Date(ie()[h+20>>>2>>>0]+1900,ie()[h+16>>>2>>>0],ie()[h+12>>>2>>>0],ie()[h+8>>>2>>>0],ie()[h+4>>>2>>>0],ie()[h>>>2>>>0],0),I=ie()[h+32>>>2>>>0],R=E.getTimezoneOffset(),H=new Date(E.getFullYear(),6,1).getTimezoneOffset(),pe=new Date(E.getFullYear(),0,1).getTimezoneOffset(),Ie=Math.min(pe,H);return 0>I?ie()[h+32>>>2>>>0]=+(H!=pe&&Ie==R):0<I!=(Ie==R)&&(H=Math.max(pe,H),E.setTime(E.getTime()+6e4*((0<I?Ie:H)-R))),ie()[h+24>>>2>>>0]=E.getDay(),I=(ko(E.getFullYear())?Io:Qn)[E.getMonth()]+E.getDate()-1|0,ie()[h+28>>>2>>>0]=I,ie()[h>>>2>>>0]=E.getSeconds(),ie()[h+4>>>2>>>0]=E.getMinutes(),ie()[h+8>>>2>>>0]=E.getHours(),ie()[h+12>>>2>>>0]=E.getDate(),ie()[h+16>>>2>>>0]=E.getMonth(),ie()[h+20>>>2>>>0]=E.getYear(),h=E.getTime(),BigInt(isNaN(h)?-1:h/1e3)}function Xn(h,E,I,R,H,pe,Ie){return c?ze(16,1,h,E,I,R,H,pe,Ie):-52}function Fo(h,E,I,R,H,pe){if(c)return ze(17,1,h,E,I,R,H,pe)}var qs={},ja=()=>performance.timeOrigin+performance.now();function Jn(h,E){if(c)return ze(18,1,h,E);if(qs[h]&&(clearTimeout(qs[h].id),delete qs[h]),!E)return 0;var I=setTimeout(()=>{delete qs[h],ue(()=>Ho(h,performance.timeOrigin+performance.now()))},E);return qs[h]={id:I,qc:E},0}function $u(h,E,I,R){h>>>=0,E>>>=0,I>>>=0,R>>>=0;var H=new Date().getFullYear(),pe=new Date(H,0,1).getTimezoneOffset();H=new Date(H,6,1).getTimezoneOffset();var Ie=Math.max(pe,H);me()[h>>>2>>>0]=60*Ie,ie()[E>>>2>>>0]=+(pe!=H),h=(E=Re=>{var Je=Math.abs(Re);return`UTC${0<=Re?"-":"+"}${String(Math.floor(Je/60)).padStart(2,"0")}${String(Je%60).padStart(2,"0")}`})(pe),E=E(H),H<pe?(Y(h,I,17),Y(E,R,17)):(Y(h,R,17),Y(E,I,17))}var Va=()=>Date.now();function Oo(h,E,I){return 0<=h&&3>=h?(h===0?h=Date.now():h=performance.timeOrigin+performance.now(),te[I>>>0>>>3]=BigInt(Math.round(1e6*h)),0):28}var Sn=[],$n=(h,E)=>{Sn.length=0;for(var I;I=W()[h++>>>0];){var R=I!=105;E+=(R&=I!=112)&&E%8?4:0,Sn.push(I==112?me()[E>>>2>>>0]:I==106?te[E>>>3]:I==105?ie()[E>>>2>>>0]:Ve()[E>>>3>>>0]),E+=R?8:4}return Sn};function Ua(h,E,I){return h>>>=0,E=$n(E>>>0,I>>>0),ce[h](...E)}function Wa(h,E,I){return h>>>=0,E=$n(E>>>0,I>>>0),ce[h](...E)}var Ga=()=>{};function Ha(h,E){return M(Kt(h>>>0,E>>>0))}var Ka=()=>{throw Ze+=1,"unwind"};function qa(){return 4294901760}var Qa=()=>navigator.hardwareConcurrency;function Xa(){return de("Cannot use emscripten_pc_get_function without -sUSE_OFFSET_CONVERTER"),0}function Do(h){h>>>=0;var E=W().length;if(h<=E||4294901760<h)return!1;for(var I=1;4>=I;I*=2){var R=E*(1+.2/I);R=Math.min(R,h+100663296);e:{R=(Math.min(4294901760,65536*Math.ceil(Math.max(h,R)/65536))-T.buffer.byteLength+65535)/65536|0;try{T.grow(R),z();var H=1;break e}catch{}H=void 0}if(H)return!0}return!1}var An=()=>(de("Cannot use convertFrameToPC (needed by __builtin_return_address) without -sUSE_OFFSET_CONVERTER"),0),Bs={},Yn=h=>{h.forEach(E=>{An()})};function Ja(){var h=Error().stack.toString().split(`
`);return h[0]=="Error"&&h.shift(),Yn(h),Bs.Lb=An(),Bs.cc=h,Bs.Lb}function Lo(h,E,I){if(h>>>=0,E>>>=0,Bs.Lb==h)var R=Bs.cc;else(R=Error().stack.toString().split(`
`))[0]=="Error"&&R.shift(),Yn(R);for(var H=3;R[H]&&An()!=h;)++H;for(h=0;h<I&&R[h+H];++h)ie()[E+4*h>>>2>>>0]=An();return h}var As,Zn={},Ya=()=>{if(!As){var h,E={USER:"web_user",LOGNAME:"web_user",PATH:"/",PWD:"/",HOME:"/home/web_user",LANG:(typeof navigator=="object"&&navigator.languages&&navigator.languages[0]||"C").replace("-","_")+".UTF-8",_:"./this.program"};for(h in Zn)Zn[h]===void 0?delete E[h]:E[h]=Zn[h];var I=[];for(h in E)I.push(`${h}=${E[h]}`);As=I}return As};function zo(h,E){if(c)return ze(19,1,h,E);h>>>=0,E>>>=0;var I=0;return Ya().forEach((R,H)=>{var pe=E+I;for(H=me()[h+4*H>>>2>>>0]=pe,pe=0;pe<R.length;++pe)O()[H++>>>0]=R.charCodeAt(pe);O()[H>>>0]=0,I+=R.length+1}),0}function Za(h,E){if(c)return ze(20,1,h,E);h>>>=0,E>>>=0;var I=Ya();me()[h>>>2>>>0]=I.length;var R=0;return I.forEach(H=>R+=H.length+1),me()[E>>>2>>>0]=R,0}function Ro(h){return c?ze(21,1,h):52}function eo(h,E,I,R){return c?ze(22,1,h,E,I,R):52}function Bo(h,E,I,R){return c?ze(23,1,h,E,I,R):70}var ei=[null,[],[]];function No(h,E,I,R){if(c)return ze(24,1,h,E,I,R);E>>>=0,I>>>=0,R>>>=0;for(var H=0,pe=0;pe<I;pe++){var Ie=me()[E>>>2>>>0],Re=me()[E+4>>>2>>>0];E+=8;for(var Je=0;Je<Re;Je++){var ot=W()[Ie+Je>>>0],Tt=ei[h];ot===0||ot===10?((h===1?x:M)(St(Tt)),Tt.length=0):Tt.push(ot)}H+=Re}return me()[R>>>2>>>0]=H,0}c||function(){for(var h=n.numThreads-1;h--;)Ar();qe.unshift(()=>{Me++,function(E){c?E():Promise.all(kt.map(sr)).then(E)}(()=>ye())})}();for(var jo=Array(256),kn=0;256>kn;++kn)jo[kn]=String.fromCharCode(kn);qr=jo,nr=n.BindingError=class extends Error{constructor(h){super(h),this.name="BindingError"}},n.InternalError=class extends Error{constructor(h){super(h),this.name="InternalError"}},Rr.push(0,1,void 0,1,null,1,!0,1,!1,1),n.count_emval_handles=()=>Rr.length/2-5-Yr.length;var ut,ti=[He,gt,ft,$,ee,oe,xe,De,nt,wt,pt,xt,tt,It,qt,Wr,Xn,Fo,Jn,zo,Za,Ro,eo,Bo,No];(async function(){function h(R,H){return ut=R.exports,ut=function(){var pe=ut,Ie={};for(let[Re,Je]of Object.entries(pe))Ie[Re]=typeof Je=="function"?(...ot)=>{ns.push(Re);try{return Je(...ot)}finally{Q||(ns.pop(),wr&&dr===1&&ns.length===0&&(dr=0,Ze+=1,gr(Qo),typeof Fibers<"u"&&Fibers.rc()))}}:Je;return Ie}(),ut=function(){var pe=ut,Ie=Je=>ot=>Je(ot)>>>0,Re=Je=>()=>Je()>>>0;return(pe=Object.assign({},pe)).Da=Ie(pe.Da),pe.fb=Re(pe.fb),pe.hb=Ie(pe.hb),pe.tb=Ie(pe.tb),pe.ub=Re(pe.ub),pe.__cxa_get_exception_ptr=Ie(pe.__cxa_get_exception_ptr),pe}(),yr.push(ut.ib),v=H,ye(),ut}Me++;var E=we();if(n.instantiateWasm)return new Promise(R=>{n.instantiateWasm(E,(H,pe)=>{h(H,pe),R(H.exports)})});if(c)return new Promise(R=>{$e=H=>{var pe=new WebAssembly.Instance(H,we());R(h(pe,H))}});Ee??(Ee=n.locateFile?n.locateFile?n.locateFile("ort-wasm-simd-threaded.jsep.wasm",g):g+"ort-wasm-simd-threaded.jsep.wasm":new URL("/assets/ort-wasm-simd-threaded.jsep-B0T3yYHD.wasm",self.location.href).href);try{var I=await async function(R){var H=Ee;if(!he&&typeof WebAssembly.instantiateStreaming=="function"&&!B(H))try{var pe=fetch(H,{credentials:"same-origin"});return await WebAssembly.instantiateStreaming(pe,R)}catch(Ie){M(`wasm streaming compile failed: ${Ie}`),M("falling back to ArrayBuffer instantiation")}return async function(Ie,Re){try{var Je=await async function(ot){if(!he)try{var Tt=await _(ot);return new Uint8Array(Tt)}catch{}if(ot==Ee&&he)ot=new Uint8Array(he);else{if(!f)throw"both async and sync fetching of the wasm failed";ot=f(ot)}return ot}(Ie);return await WebAssembly.instantiate(Je,Re)}catch(ot){M(`failed to asynchronously prepare wasm: ${ot}`),de(ot)}}(H,R)}(E);return h(I.instance,I.module)}catch(R){return o(R),Promise.reject(R)}})();var Vo=h=>(Vo=ut.Da)(h),Uo=()=>(Uo=ut.Ea)();n._OrtInit=(h,E)=>(n._OrtInit=ut.Fa)(h,E),n._OrtGetLastError=(h,E)=>(n._OrtGetLastError=ut.Ga)(h,E),n._OrtCreateSessionOptions=(h,E,I,R,H,pe,Ie,Re,Je,ot)=>(n._OrtCreateSessionOptions=ut.Ha)(h,E,I,R,H,pe,Ie,Re,Je,ot),n._OrtAppendExecutionProvider=(h,E,I,R,H)=>(n._OrtAppendExecutionProvider=ut.Ia)(h,E,I,R,H),n._OrtAddFreeDimensionOverride=(h,E,I)=>(n._OrtAddFreeDimensionOverride=ut.Ja)(h,E,I),n._OrtAddSessionConfigEntry=(h,E,I)=>(n._OrtAddSessionConfigEntry=ut.Ka)(h,E,I),n._OrtReleaseSessionOptions=h=>(n._OrtReleaseSessionOptions=ut.La)(h),n._OrtCreateSession=(h,E,I)=>(n._OrtCreateSession=ut.Ma)(h,E,I),n._OrtReleaseSession=h=>(n._OrtReleaseSession=ut.Na)(h),n._OrtGetInputOutputCount=(h,E,I)=>(n._OrtGetInputOutputCount=ut.Oa)(h,E,I),n._OrtGetInputOutputMetadata=(h,E,I,R)=>(n._OrtGetInputOutputMetadata=ut.Pa)(h,E,I,R),n._OrtFree=h=>(n._OrtFree=ut.Qa)(h),n._OrtCreateTensor=(h,E,I,R,H,pe)=>(n._OrtCreateTensor=ut.Ra)(h,E,I,R,H,pe),n._OrtGetTensorData=(h,E,I,R,H)=>(n._OrtGetTensorData=ut.Sa)(h,E,I,R,H),n._OrtReleaseTensor=h=>(n._OrtReleaseTensor=ut.Ta)(h),n._OrtCreateRunOptions=(h,E,I,R)=>(n._OrtCreateRunOptions=ut.Ua)(h,E,I,R),n._OrtAddRunConfigEntry=(h,E,I)=>(n._OrtAddRunConfigEntry=ut.Va)(h,E,I),n._OrtReleaseRunOptions=h=>(n._OrtReleaseRunOptions=ut.Wa)(h),n._OrtCreateBinding=h=>(n._OrtCreateBinding=ut.Xa)(h),n._OrtBindInput=(h,E,I)=>(n._OrtBindInput=ut.Ya)(h,E,I),n._OrtBindOutput=(h,E,I,R)=>(n._OrtBindOutput=ut.Za)(h,E,I,R),n._OrtClearBoundOutputs=h=>(n._OrtClearBoundOutputs=ut._a)(h),n._OrtReleaseBinding=h=>(n._OrtReleaseBinding=ut.$a)(h),n._OrtRunWithBinding=(h,E,I,R,H)=>(n._OrtRunWithBinding=ut.ab)(h,E,I,R,H),n._OrtRun=(h,E,I,R,H,pe,Ie,Re)=>(n._OrtRun=ut.bb)(h,E,I,R,H,pe,Ie,Re),n._OrtEndProfiling=h=>(n._OrtEndProfiling=ut.cb)(h),n._JsepOutput=(h,E,I)=>(n._JsepOutput=ut.db)(h,E,I),n._JsepGetNodeName=h=>(n._JsepGetNodeName=ut.eb)(h);var ln=()=>(ln=ut.fb)(),os=n._free=h=>(os=n._free=ut.gb)(h),In=n._malloc=h=>(In=n._malloc=ut.hb)(h),Fn=(h,E,I,R,H,pe)=>(Fn=ut.kb)(h,E,I,R,H,pe),Wo=()=>(Wo=ut.lb)(),Go=(h,E,I,R,H)=>(Go=ut.mb)(h,E,I,R,H),to=h=>(to=ut.nb)(h),ro=h=>(ro=ut.ob)(h),Ho=(h,E)=>(Ho=ut.pb)(h,E),so=()=>(so=ut.qb)(),Ko=(h,E)=>(Ko=ut.rb)(h,E),On=h=>(On=ut.sb)(h),Dn=h=>(Dn=ut.tb)(h),no=()=>(no=ut.ub)(),qo=n.dynCall_ii=(h,E)=>(qo=n.dynCall_ii=ut.vb)(h,E),oo=h=>(oo=ut.wb)(h),Qo=()=>(Qo=ut.xb)(),Xo=h=>(Xo=ut.yb)(h),ao=()=>(ao=ut.zb)();return n.stackSave=()=>no(),n.stackRestore=h=>On(h),n.stackAlloc=h=>Dn(h),n.setValue=function(h,E,I="i8"){switch(I.endsWith("*")&&(I="*"),I){case"i1":case"i8":O()[h>>>0]=E;break;case"i16":N()[h>>>1>>>0]=E;break;case"i32":ie()[h>>>2>>>0]=E;break;case"i64":te[h>>>3]=BigInt(E);break;case"float":Ae()[h>>>2>>>0]=E;break;case"double":Ve()[h>>>3>>>0]=E;break;case"*":me()[h>>>2>>>0]=E;break;default:de(`invalid type for setValue: ${I}`)}},n.getValue=function(h,E="i8"){switch(E.endsWith("*")&&(E="*"),E){case"i1":case"i8":return O()[h>>>0];case"i16":return N()[h>>>1>>>0];case"i32":return ie()[h>>>2>>>0];case"i64":return te[h>>>3];case"float":return Ae()[h>>>2>>>0];case"double":return Ve()[h>>>3>>>0];case"*":return me()[h>>>2>>>0];default:de(`invalid type for getValue: ${E}`)}},n.UTF8ToString=Kt,n.stringToUTF8=Y,n.lengthBytesUTF8=V,function h(){if(0<Me)Ce=h;else if(c)s(n),_e();else{for(;0<qe.length;)qe.shift()(n);0<Me?Ce=h:(n.calledRun=!0,Q||(_e(),s(n)))}}(),n.PTR_SIZE=4,a}),Ed=Bi,Pd=(r=(e=globalThis.self)==null?void 0:e.name)==null?void 0:r.startsWith("em-pthread"),Pd&&Bi()}),Ni,ji,Cd,ts,Sd,na,$d,Ad,Vi,kd,Ui,Id,Wi,Fd,Gi=Ue(()=>{Di(),Ni=typeof location>"u"?void 0:location.origin,ji=self.location.href>"file:"&&self.location.href<"file;",Cd=()=>{{if(ji){let e=URL;return new URL(new e("ort.bundle.min.mjs",self.location.href).href,Ni).href}return self.location.href}},ts=Cd(),Sd=()=>{if(ts&&!ts.startsWith("blob:"))return ts.substring(0,ts.lastIndexOf("/")+1)},na=(e,r)=>{try{let t=r??ts;return(t?new URL(e,t):new URL(e)).origin===Ni}catch{return!1}},$d=(e,r)=>{let t=r??ts;try{return(t?new URL(e,t):new URL(e)).href}catch{return}},Ad=(e,r)=>`${r??"./"}${e}`,Vi=async e=>{let r=await(await fetch(e,{credentials:"same-origin"})).blob();return URL.createObjectURL(r)},kd=async e=>(await import(e)).default,Ui=(jx(),co(vd)).default,Id=async()=>{if(!ts)throw new Error("Failed to load proxy worker: cannot determine the script source URL.");if(na(ts))return[void 0,Ui()];let e=await Vi(ts);return[e,Ui(e)]},Wi=(Vx(),co(Td)).default,Fd=async(e,r,t)=>{if(!e&&!r&&Wi&&ts&&na(ts))return[void 0,Wi];{let s="ort-wasm-simd-threaded.jsep.mjs",o=e??$d(s,r),n=t&&o&&!na(o,r),a=n?await Vi(o):o??Ad(s,r);return[n?a:void 0,await kd(a)]}}}),Hi,oa,fo,Ki,Od,Dd,Ld,qi,Xt,mn=Ue(()=>{Gi(),oa=!1,fo=!1,Ki=!1,Od=()=>{if(typeof SharedArrayBuffer>"u")return!1;try{return typeof MessageChannel<"u"&&new MessageChannel().port1.postMessage(new SharedArrayBuffer(1)),WebAssembly.validate(new Uint8Array([0,97,115,109,1,0,0,0,1,4,1,96,0,0,3,2,1,0,5,4,1,3,1,1,10,11,1,9,0,65,0,254,16,2,0,26,11]))}catch{return!1}},Dd=()=>{try{return WebAssembly.validate(new Uint8Array([0,97,115,109,1,0,0,0,1,4,1,96,0,0,3,2,1,0,10,30,1,28,0,65,0,253,15,253,12,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,253,186,1,26,11]))}catch{return!1}},Ld=()=>{try{return WebAssembly.validate(new Uint8Array([0,97,115,109,1,0,0,0,1,5,1,96,0,1,123,3,2,1,0,10,19,1,17,0,65,1,253,15,65,2,253,15,65,3,253,15,253,147,2,11]))}catch{return!1}},qi=async e=>{if(oa)return Promise.resolve();if(fo)throw new Error("multiple calls to 'initializeWebAssembly()' detected.");if(Ki)throw new Error("previous call to 'initializeWebAssembly()' failed.");fo=!0;let r=e.initTimeout,t=e.numThreads;if(e.simd!==!1){if(e.simd==="relaxed"){if(!Ld())throw new Error("Relaxed WebAssembly SIMD is not supported in the current environment.")}else if(!Dd())throw new Error("WebAssembly SIMD is not supported in the current environment.")}let s=Od();t>1&&!s&&(typeof self<"u"&&!self.crossOriginIsolated&&console.warn("env.wasm.numThreads is set to "+t+", but this will not work unless you enable crossOriginIsolated mode. See https://web.dev/cross-origin-isolation-guide/ for more info."),console.warn("WebAssembly multi-threading is not supported in the current environment. Falling back to single-threading."),e.numThreads=t=1);let o=e.wasmPaths,n=typeof o=="string"?o:void 0,a=o==null?void 0:o.mjs,i=(a==null?void 0:a.href)??a,l=o==null?void 0:o.wasm,c=(l==null?void 0:l.href)??l,p=e.wasmBinary,[d,u]=await Fd(i,n,t>1),_=!1,f=[];if(r>0&&f.push(new Promise(b=>{setTimeout(()=>{_=!0,b()},r)})),f.push(new Promise((b,A)=>{let g={numThreads:t};if(p)g.wasmBinary=p;else if(c||n)g.locateFile=y=>c??n+y;else if(i&&i.indexOf("blob:")!==0)g.locateFile=y=>new URL(y,i).href;else if(d){let y=Sd();y&&(g.locateFile=C=>y+C)}u(g).then(y=>{fo=!1,oa=!0,Hi=y,b(),d&&URL.revokeObjectURL(d)},y=>{fo=!1,Ki=!0,A(y)})})),await Promise.race(f),_)throw new Error(`WebAssembly backend initializing failed due to timeout: ${r}ms`)},Xt=()=>{if(oa&&Hi)return Hi;throw new Error("WebAssembly is not initialized yet.")}}),ys,aa,Wt,Qi=Ue(()=>{mn(),ys=(e,r)=>{let t=Xt(),s=t.lengthBytesUTF8(e)+1,o=t._malloc(s);return t.stringToUTF8(e,o,s),r.push(o),o},aa=(e,r,t,s)=>{if(typeof e=="object"&&e!==null){if(t.has(e))throw new Error("Circular reference in options");t.add(e)}Object.entries(e).forEach(([o,n])=>{let a=r?r+o:o;if(typeof n=="object")aa(n,a+".",t,s);else if(typeof n=="string"||typeof n=="number")s(a,n.toString());else if(typeof n=="boolean")s(a,n?"1":"0");else throw new Error(`Can't handle extra config type: ${typeof n}`)})},Wt=e=>{let r=Xt(),t=r.stackSave();try{let s=r.PTR_SIZE,o=r.stackAlloc(2*s);r._OrtGetLastError(o,o+s);let n=Number(r.getValue(o,s===4?"i32":"i64")),a=r.getValue(o+s,"*"),i=a?r.UTF8ToString(a):"";throw new Error(`${e} ERROR_CODE: ${n}, ERROR_MESSAGE: ${i}`)}finally{r.stackRestore(t)}}}),zd,Ux=Ue(()=>{mn(),Qi(),zd=e=>{let r=Xt(),t=0,s=[],o=e||{};try{if((e==null?void 0:e.logSeverityLevel)===void 0)o.logSeverityLevel=2;else if(typeof e.logSeverityLevel!="number"||!Number.isInteger(e.logSeverityLevel)||e.logSeverityLevel<0||e.logSeverityLevel>4)throw new Error(`log serverity level is not valid: ${e.logSeverityLevel}`);if((e==null?void 0:e.logVerbosityLevel)===void 0)o.logVerbosityLevel=0;else if(typeof e.logVerbosityLevel!="number"||!Number.isInteger(e.logVerbosityLevel))throw new Error(`log verbosity level is not valid: ${e.logVerbosityLevel}`);(e==null?void 0:e.terminate)===void 0&&(o.terminate=!1);let n=0;return(e==null?void 0:e.tag)!==void 0&&(n=ys(e.tag,s)),t=r._OrtCreateRunOptions(o.logSeverityLevel,o.logVerbosityLevel,!!o.terminate,n),t===0&&Wt("Can't create run options."),(e==null?void 0:e.extra)!==void 0&&aa(e.extra,"",new WeakSet,(a,i)=>{let l=ys(a,s),c=ys(i,s);r._OrtAddRunConfigEntry(t,l,c)!==0&&Wt(`Can't set a run config entry: ${a} - ${i}.`)}),[t,s]}catch(n){throw t!==0&&r._OrtReleaseRunOptions(t),s.forEach(a=>r._free(a)),n}}}),Rd,Bd,Nd,_o,jd,Vd,Wx=Ue(()=>{mn(),Qi(),Rd=e=>{switch(e){case"disabled":return 0;case"basic":return 1;case"extended":return 2;case"all":return 99;default:throw new Error(`unsupported graph optimization level: ${e}`)}},Bd=e=>{switch(e){case"sequential":return 0;case"parallel":return 1;default:throw new Error(`unsupported execution mode: ${e}`)}},Nd=e=>{e.extra||(e.extra={}),e.extra.session||(e.extra.session={});let r=e.extra.session;r.use_ort_model_bytes_directly||(r.use_ort_model_bytes_directly="1"),e.executionProviders&&e.executionProviders.some(t=>(typeof t=="string"?t:t.name)==="webgpu")&&(e.enableMemPattern=!1)},_o=(e,r,t,s)=>{let o=ys(r,s),n=ys(t,s);Xt()._OrtAddSessionConfigEntry(e,o,n)!==0&&Wt(`Can't set a session config entry: ${r} - ${t}.`)},jd=async(e,r,t)=>{for(let s of r){let o=typeof s=="string"?s:s.name,n=[];switch(o){case"webnn":if(o="WEBNN",typeof s!="string"){let p=s==null?void 0:s.deviceType;p&&_o(e,"deviceType",p,t)}break;case"webgpu":if(o="JS",typeof s!="string"){let p=s;if(p!=null&&p.preferredLayout){if(p.preferredLayout!=="NCHW"&&p.preferredLayout!=="NHWC")throw new Error(`preferredLayout must be either 'NCHW' or 'NHWC': ${p.preferredLayout}`);_o(e,"preferredLayout",p.preferredLayout,t)}}break;case"wasm":case"cpu":continue;default:throw new Error(`not supported execution provider: ${o}`)}let a=ys(o,t),i=n.length,l=0,c=0;if(i>0){l=Xt()._malloc(i*Xt().PTR_SIZE),t.push(l),c=Xt()._malloc(i*Xt().PTR_SIZE),t.push(c);for(let p=0;p<i;p++)Xt().setValue(l+p*Xt().PTR_SIZE,n[p][0],"*"),Xt().setValue(c+p*Xt().PTR_SIZE,n[p][1],"*")}await Xt()._OrtAppendExecutionProvider(e,a,l,c,i)!==0&&Wt(`Can't append execution provider: ${o}.`)}},Vd=async e=>{let r=Xt(),t=0,s=[],o=e||{};Nd(o);try{let n=Rd(o.graphOptimizationLevel??"all"),a=Bd(o.executionMode??"sequential"),i=typeof o.logId=="string"?ys(o.logId,s):0,l=o.logSeverityLevel??2;if(!Number.isInteger(l)||l<0||l>4)throw new Error(`log serverity level is not valid: ${l}`);let c=o.logVerbosityLevel??0;if(!Number.isInteger(c)||c<0||c>4)throw new Error(`log verbosity level is not valid: ${c}`);let p=typeof o.optimizedModelFilePath=="string"?ys(o.optimizedModelFilePath,s):0;if(t=r._OrtCreateSessionOptions(n,!!o.enableCpuMemArena,!!o.enableMemPattern,a,!!o.enableProfiling,0,i,l,c,p),t===0&&Wt("Can't create session options."),o.executionProviders&&await jd(t,o.executionProviders,s),o.enableGraphCapture!==void 0){if(typeof o.enableGraphCapture!="boolean")throw new Error(`enableGraphCapture must be a boolean value: ${o.enableGraphCapture}`);_o(t,"enableGraphCapture",o.enableGraphCapture.toString(),s)}if(o.freeDimensionOverrides)for(let[d,u]of Object.entries(o.freeDimensionOverrides)){if(typeof d!="string")throw new Error(`free dimension override name must be a string: ${d}`);if(typeof u!="number"||!Number.isInteger(u)||u<0)throw new Error(`free dimension override value must be a non-negative integer: ${u}`);let _=ys(d,s);r._OrtAddFreeDimensionOverride(t,_,u)!==0&&Wt(`Can't set a free dimension override: ${d} - ${u}.`)}return o.extra!==void 0&&aa(o.extra,"",new WeakSet,(d,u)=>{_o(t,d,u,s)}),[t,s]}catch(n){throw t!==0&&r._OrtReleaseSessionOptions(t)!==0&&Wt("Can't release session options."),s.forEach(a=>r._free(a)),n}}}),Wn,Ns,fn,Xi,ia,Ji,Yi,Zi,Mt=Ue(()=>{Wn=e=>{switch(e){case"int8":return 3;case"uint8":return 2;case"bool":return 9;case"int16":return 5;case"uint16":return 4;case"int32":return 6;case"uint32":return 12;case"float16":return 10;case"float32":return 1;case"float64":return 11;case"string":return 8;case"int64":return 7;case"uint64":return 13;case"int4":return 22;case"uint4":return 21;default:throw new Error(`unsupported data type: ${e}`)}},Ns=e=>{switch(e){case 3:return"int8";case 2:return"uint8";case 9:return"bool";case 5:return"int16";case 4:return"uint16";case 6:return"int32";case 12:return"uint32";case 10:return"float16";case 1:return"float32";case 11:return"float64";case 8:return"string";case 7:return"int64";case 13:return"uint64";case 22:return"int4";case 21:return"uint4";default:throw new Error(`unsupported data type: ${e}`)}},fn=(e,r)=>{let t=[-1,4,1,1,2,2,4,8,-1,1,2,8,4,8,-1,-1,-1,-1,-1,-1,-1,.5,.5][e],s=typeof r=="number"?r:r.reduce((o,n)=>o*n,1);return t>0?Math.ceil(s*t):void 0},Xi=e=>{switch(e){case"float16":return typeof Float16Array<"u"&&Float16Array.from?Float16Array:Uint16Array;case"float32":return Float32Array;case"uint8":return Uint8Array;case"int8":return Int8Array;case"uint16":return Uint16Array;case"int16":return Int16Array;case"int32":return Int32Array;case"bool":return Uint8Array;case"float64":return Float64Array;case"uint32":return Uint32Array;case"int64":return BigInt64Array;case"uint64":return BigUint64Array;default:throw new Error(`unsupported type: ${e}`)}},ia=e=>{switch(e){case"verbose":return 0;case"info":return 1;case"warning":return 2;case"error":return 3;case"fatal":return 4;default:throw new Error(`unsupported logging level: ${e}`)}},Ji=e=>e==="float32"||e==="float16"||e==="int32"||e==="int64"||e==="uint32"||e==="uint8"||e==="bool"||e==="uint4"||e==="int4",Yi=e=>e==="float32"||e==="float16"||e==="int32"||e==="int64"||e==="uint32"||e==="uint64"||e==="int8"||e==="uint8"||e==="bool"||e==="uint4"||e==="int4",Zi=e=>{switch(e){case"none":return 0;case"cpu":return 1;case"cpu-pinned":return 2;case"texture":return 3;case"gpu-buffer":return 4;case"ml-tensor":return 5;default:throw new Error(`unsupported data location: ${e}`)}}}),el,Ud=Ue(()=>{Di(),el=async e=>{if(typeof e=="string"){let r=await fetch(e);if(!r.ok)throw new Error(`failed to load external data file: ${e}`);let t=r.headers.get("Content-Length"),s=t?parseInt(t,10):0;if(s<1073741824)return new Uint8Array(await r.arrayBuffer());{if(!r.body)throw new Error(`failed to load external data file: ${e}, no response body.`);let o=r.body.getReader(),n;try{n=new ArrayBuffer(s)}catch(i){if(i instanceof RangeError){let l=Math.ceil(s/65536);n=new WebAssembly.Memory({initial:l,maximum:l}).buffer}else throw i}let a=0;for(;;){let{done:i,value:l}=await o.read();if(i)break;let c=l.byteLength;new Uint8Array(n,a,c).set(l),a+=c}return new Uint8Array(n,0,s)}}else return e instanceof Blob?new Uint8Array(await e.arrayBuffer()):e instanceof Uint8Array?e:new Uint8Array(e)}}),Wd,Gd,Hd,Kd,tl,qd,Dt,js=Ue(()=>{Mt(),Wd=["V","I","W","E","F"],Gd=(e,r)=>{console.log(`[${Wd[e]},${new Date().toISOString()}]${r}`)},tl=(e,r)=>{Hd=e,Kd=r},qd=(e,r)=>{let t=ia(e),s=ia(Hd);t>=s&&Gd(t,typeof r=="function"?r():r)},Dt=(...e)=>{Kd&&qd(...e)}}),Qd,Gn,Pe,la,Xd,Jd,Yd,Pt=Ue(()=>{Qd=class{static calcMatMulShape(e,r){return e[1]!==r[0]?void 0:[e[0],r[1]]}},Gn=class{static calcShape(e,r,t=!1){let s=e.length,o=r.length;if(s===0)return r;if(o===0)return e;let n=Math.max(e.length,r.length),a=new Array(n);if(t){if(s<2||o<2)return;let i=Qd.calcMatMulShape([e[s-2],e[s-1]],[r[o-2],r[o-1]]);if(i===void 0)return;[a[n-2],a[n-1]]=i}for(let i=t?3:1;i<=n;i++){let l=s-i<0?1:e[s-i],c=o-i<0?1:r[o-i];if(l!==c&&l>1&&c>1)return;let p=Math.max(l,c);if(l&&c)a[n-i]=Math.max(l,c);else{if(p>1)return;a[n-i]=0}}return a}static isValidBroadcast(e,r){let t=e.length,s=r.length;if(t>s)return!1;for(let o=1;o<=t;o++)if(e[t-o]!==1&&e[t-o]!==r[s-o])return!1;return!0}},Pe=class xi{static size(r){return xi.getSizeFromDimensionRange(r,0,r.length)}static convertShape(r,t=4){let s=r.length;if(s===0)return[];let o=new Array(s),n=s-1;for(;n>=0;){if(r[n]%t===0){o[n]=r[n]/t;break}if(t%r[n]!==0)throw new Error("cannot convert shape");o[n]=1,t/=r[n],n--}for(n--;n>=0;n--)o[n]=r[n];return o}static sizeFromDimension(r,t){if(t<0||t>r.length)throw new Error(`invalid dimension of ${t} for sizeFromDimension as Tensor has ${r.length} dimensions.`);return xi.getSizeFromDimensionRange(r,t,r.length)}static sizeToDimension(r,t){if(t<0||t>r.length)throw new Error(`invalid dimension of ${t} for sizeToDimension as Tensor has ${r.length} dimensions.`);return xi.getSizeFromDimensionRange(r,0,t)}static getSizeFromDimensionRange(r,t,s){let o=1;for(let n=t;n<s;n++){if(r[n]<0)throw new Error("cannot get valid size from specified dimension range. Most likely the range contains negative values in them.");o*=Number(r[n])}return o}static computeStrides(r){let t=r.length;if(t===0)return[];if(t===1)return[1];let s=new Array(t);s[t-1]=1,s[t-2]=r[t-1];for(let o=t-3;o>=0;--o)s[o]=s[o+1]*r[o+1];return s}static normalizeAxis(r,t){if(r<-t&&r>=t)throw new Error("unsupported axis for this operation.");return r<0?r+t:r}static normalizeAxes(r,t){return r.map(s=>this.normalizeAxis(s,t??r.length))}static sortBasedOnPerm(r,t){return t?t.map(s=>r[s]):r.slice().reverse()}static padShape(r,t){let s=r.length;return r.map((o,n)=>o+t[n]+t[n+s])}static areEqual(r,t){return r.length!==t.length?!1:r.every((s,o)=>s===t[o])}},la=class ta{static adjustPoolAttributes(r,t,s,o,n,a){if(!r&&s.length!==t.length-2)throw new Error("length of specified kernel shapes should be 2 less than length of input dimensions");if(r)for(let i=0;i<t.length-2;i++)i>=s.length?s.push(t[i+2]):s[i]=t[i+2];for(let i=0;i<s.length;i++)if(i<o.length){if(o[i]<0)throw new Error("strides should be greater than or equal to 1")}else o.push(1);for(let i=0;i<s.length;i++)if(i<n.length){if(n[i]<0)throw new Error("dilations should be greater than or equal to 1")}else n.push(1);for(let i=0;i<s.length*2;i++)if(i<a.length){if(a[i]<0)throw new Error("pad should be greater than or equal to 1")}else a.push(0);for(let i=0;i<s.length;i++){if(s[i]<=0)throw new Error("kernel shapes need to be greater than 0");if(a[i]>=s[i]||a[i+s.length]>=s[i])throw new Error("pads should be smaller than kernel")}}static adjustPadsBasedOnAutoPad(r,t,s,o,n,a,i){if(i){if(n.length!==2*(r.length-2))throw new Error("length of pads should be twice the length of data dimensions");if(t.length!==r.length-2)throw new Error("length of strides should be the length of data dimensions");if(o.length!==r.length-2)throw new Error("length of kernel shapes should be the length of data dimensions");for(let l=0;l<r.length-2;l++)ta.adjustPadAndReturnShape(r[l+(a?1:2)],t[l],s[l],o[l],n,l,l+r.length-2,i)}}static computePoolOutputShape(r,t,s,o,n,a,i){if(t.length<=0)throw new Error("input shape must be of size greater than 0");let l=[t[0],t[1]];return ta.computeShapeHelper(r,t,l,s,o,n,a,i),l}static computeConvOutputShape(r,t,s,o,n,a,i){if(r.length<=0||t.length<=0)throw new Error("invalid input tensor dims or invalid filter tensor dims");let l=[r[0],t[0]];return ta.computeShapeHelper(!1,r,l,s,o,n,a,i),l}static computeShapeHelper(r,t,s,o,n,a,i,l){if(r)for(let c=0;c<t.length-2;c++)s.push(1);else for(let c=0;c<t.length-2;c++)s.push(ta.adjustPadAndReturnShape(t[c+2],o[c],n[c],a[c],i,c,c+t.length-2,l))}static adjustPadAndReturnShape(r,t,s,o,n,a,i,l){let c=s*(o-1)+1;if(l&&l!=="NOTSET")switch(l){case"VALID":return n[a]=0,n[i]=0,Math.floor((r-c)/t+1);case"SAME_LOWER":case"SAME_UPPER":if(s!==1)throw new Error("Dilation not supported for SAME_UPPER or SAME_LOWER");{let p=((r+t-1)/t-1)*t+o-r;return n[a]=Math.floor(l==="SAME_LOWER"?(p+1)/2:p/2),n[i]=p-n[a],Math.floor((r+p-o)/t+1)}default:throw new Error("Unsupported AutoPad type")}else return Math.floor((r+n[a]+n[i]-c)/t+1)}},Xd=class{static getShapeOfGemmResult(e,r,t,s,o){if(e.length!==2||t.length!==2)throw new Error("shape need to be of size 2");let n,a,i;r?(n=e[1],a=e[0]):(n=e[0],a=e[1]);let l=-1;if(s?(i=t[0],l=1):(i=t[1],l=0),t[l]!==a)throw new Error("dimension mismatch");if(n<=0||i<=0||a<=0)throw new Error("invalid shape specified");if(o&&!Gn.isValidBroadcast(o,[n,i]))throw new Error("gemm: invalid bias shape for broadcast");return[n,i,a]}},Jd=-34028234663852886e22,Yd=34028234663852886e22}),rl,Zd=Ue(()=>{Mt(),rl=(e,r)=>new(Xi(r))(e)}),sl,nl,ep,ol,tp,al,il,ll,rp,sp,Gx=Ue(()=>{js(),sl=(e,r=!0)=>{if(e.byteLength%8!==0)throw new Error("Invalid Uint8Array length - must be a multiple of 8 (BigInt).");let t=e.byteLength/8,s=new BigInt64Array(e.buffer,e.byteOffset,t),o=new Int32Array(t);for(let n=0;n<t;n++){let a=s[n];if(a>2147483647n||a<-2147483648n)throw new Error(`Overflow occurred when converting BigInt to Int32 at index ${n}: ${a}`);o[n]=Number(a)}return r?new Uint8Array(o.buffer):o},nl=(e,r=!0)=>{if(e.byteLength%4!==0)throw new Error("Invalid Uint8Array length - must be a multiple of 4 (Int32).");let t=e.byteLength/4,s=new Int32Array(e.buffer,e.byteOffset,t),o=BigInt64Array.from(s,BigInt);return r?new Uint8Array(o.buffer):o},ep=1,ol=()=>ep++,tp=new Map([["float32",32],["float16",16],["int32",32],["uint32",32],["int64",64],["uint64",64],["int8",8],["uint8",8],["int4",4],["uint4",4]]),al=(e,r)=>{let t=tp.get(e);if(!t)throw new Error("Unsupported data type.");return r.length>0?Math.ceil(r.reduce((s,o)=>s*o)*t/8):0},il=class{constructor(e){this.shouldConvertInt64toInt32=!1,this.isInt64ToInt32Converted=!1;let{sessionId:r,context:t,tensor:s,dataType:o,shape:n,shouldConvertInt64toInt32:a=!1}=e;this.sessionId=r,this.mlContext=t,this.mlTensor=s,this.dataType=o,this.tensorShape=n,this.shouldConvertInt64toInt32=a}get tensor(){return this.mlTensor}get type(){return this.dataType}get shape(){return this.tensorShape}get byteLength(){return al(this.dataType,this.tensorShape)}destroy(){Dt("verbose",()=>"[WebNN] TensorWrapper.destroy"),this.mlTensor.destroy()}write(e){this.mlContext.writeTensor(this.mlTensor,e)}async read(e,r){if(e){let t=await this.mlContext.readTensor(this.mlTensor),s=nl(new Uint8Array(t));if(r){(r instanceof ArrayBuffer?new Uint8Array(r):new Uint8Array(r.buffer,r.byteOffset,r.byteLength)).set(s);return}else return s.buffer}else return r?this.mlContext.readTensor(this.mlTensor,r):this.mlContext.readTensor(this.mlTensor)}canReuseTensor(e,r,t){return this.mlContext===e&&this.dataType===r&&this.tensorShape.length===t.length&&this.tensorShape.every((s,o)=>s===t[o])}setIsInt64ToInt32Converted(e){this.isInt64ToInt32Converted=e}},ll=class{constructor(e,r){this.tensorManager=e,this.wrapper=r}get tensorWrapper(){return this.wrapper}releaseTensor(){this.tensorWrapper&&(this.tensorManager.releaseTensor(this.tensorWrapper),this.wrapper=void 0)}async ensureTensor(e,r,t,s){let o=r,n=this.tensorManager.getMLContext(e),a=o==="int64"&&!n.opSupportLimits().input.dataTypes.includes("int64");if(a&&(o="int32",Dt("verbose",()=>"[WebNN] TensorIdTracker.ensureTensor: convert dataType from int64 to int32")),this.wrapper){if(this.wrapper.canReuseTensor(n,o,t))return this.wrapper.tensor;if(s){if(this.wrapper.byteLength!==al(o,t))throw new Error("Unable to copy data to tensor with different size.");this.activeUpload=new Uint8Array(await this.wrapper.read())}this.tensorManager.releaseTensor(this.wrapper)}let i=typeof MLTensorUsage>"u"?void 0:MLTensorUsage.READ|MLTensorUsage.WRITE;return this.wrapper=await this.tensorManager.getCachedTensor(e,o,t,i,!0,!0,a),s&&this.activeUpload&&(this.wrapper.write(this.activeUpload),this.activeUpload=void 0),this.wrapper.tensor}upload(e){let r=e;if(this.wrapper)if(this.wrapper.shouldConvertInt64toInt32&&(r=sl(e,!0),this.wrapper.setIsInt64ToInt32Converted(!0)),r.byteLength===this.wrapper.byteLength){this.wrapper.write(r);return}else Dt("verbose",()=>"Data size does not match tensor size. Releasing tensor."),this.releaseTensor();this.activeUpload?this.activeUpload.set(r):this.activeUpload=new Uint8Array(r)}async download(e){var r,t,s;if(this.activeUpload){let o=(r=this.wrapper)!=null&&r.isInt64ToInt32Converted?nl(this.activeUpload):this.activeUpload;if(e){e instanceof ArrayBuffer?new Uint8Array(e).set(o):new Uint8Array(e.buffer,e.byteOffset,e.byteLength).set(o);return}else return o.buffer}if(!this.wrapper)throw new Error("Tensor has not been created.");return e?this.wrapper.read((t=this.wrapper)==null?void 0:t.shouldConvertInt64toInt32,e):this.wrapper.read((s=this.wrapper)==null?void 0:s.shouldConvertInt64toInt32)}},rp=class{constructor(e){this.backend=e,this.tensorTrackersById=new Map,this.freeTensors=[],this.externalTensors=new Set}getMLContext(e){let r=this.backend.getMLContext(e);if(!r)throw new Error("MLContext not found for session.");return r}reserveTensorId(){let e=ol();return this.tensorTrackersById.set(e,new ll(this)),e}releaseTensorId(e){let r=this.tensorTrackersById.get(e);r&&(this.tensorTrackersById.delete(e),r.tensorWrapper&&this.releaseTensor(r.tensorWrapper))}async ensureTensor(e,r,t,s,o){Dt("verbose",()=>`[WebNN] TensorManager.ensureTensor {tensorId: ${r}, dataType: ${t}, shape: ${s}, copyOld: ${o}}`);let n=this.tensorTrackersById.get(r);if(!n)throw new Error("Tensor not found.");return n.ensureTensor(e,t,s,o)}upload(e,r){let t=this.tensorTrackersById.get(e);if(!t)throw new Error("Tensor not found.");t.upload(r)}async download(e,r){Dt("verbose",()=>`[WebNN] TensorManager.download {tensorId: ${e}, dstBuffer: ${r==null?void 0:r.byteLength}}`);let t=this.tensorTrackersById.get(e);if(!t)throw new Error("Tensor not found.");return t.download(r)}releaseTensorsForSession(e){for(let r of this.freeTensors)r.sessionId===e&&r.destroy();this.freeTensors=this.freeTensors.filter(r=>r.sessionId!==e)}registerTensor(e,r,t,s){let o=this.getMLContext(e),n=ol(),a=new il({sessionId:e,context:o,tensor:r,dataType:t,shape:s});return this.tensorTrackersById.set(n,new ll(this,a)),this.externalTensors.add(a),n}async getCachedTensor(e,r,t,s,o,n,a=!1){let i=this.getMLContext(e);for(let[c,p]of this.freeTensors.entries())if(p.canReuseTensor(i,r,t)){Dt("verbose",()=>`[WebNN] Reusing tensor {dataType: ${r}, shape: ${t}}`);let d=this.freeTensors.splice(c,1)[0];return d.sessionId=e,d}Dt("verbose",()=>`[WebNN] MLContext.createTensor {dataType: ${r}, shape: ${t}}`);let l=await i.createTensor({dataType:r,shape:t,dimensions:t,usage:s,writable:o,readable:n});return new il({sessionId:e,context:i,tensor:l,dataType:r,shape:t,shouldConvertInt64toInt32:a})}releaseTensor(e){this.externalTensors.has(e)&&this.externalTensors.delete(e),this.freeTensors.push(e)}},sp=(...e)=>new rp(...e)}),ua,np,op,Hx=Ue(()=>{Mt(),mn(),Zd(),Gx(),js(),ua=new Map([[1,"float32"],[10,"float16"],[6,"int32"],[12,"uint32"],[7,"int64"],[13,"uint64"],[22,"int4"],[21,"uint4"],[3,"int8"],[2,"uint8"],[9,"uint8"]]),np=(e,r)=>{if(e===r)return!0;if(e===void 0||r===void 0)return!1;let t=Object.keys(e).sort(),s=Object.keys(r).sort();return t.length===s.length&&t.every((o,n)=>o===s[n]&&e[o]===r[o])},op=class{constructor(e){this.tensorManager=sp(this),this.mlContextBySessionId=new Map,this.sessionIdsByMLContext=new Map,this.mlContextCache=[],this.sessionGraphInputs=new Map,this.temporaryGraphInputs=[],this.temporarySessionTensorIds=new Map,tl(e.logLevel,!!e.debug)}get currentSessionId(){if(this.activeSessionId===void 0)throw new Error("No active session");return this.activeSessionId}onRunStart(e){Dt("verbose",()=>`[WebNN] onRunStart {sessionId: ${e}}`),this.activeSessionId=e}onRunEnd(e){Dt("verbose",()=>`[WebNN] onRunEnd {sessionId: ${e}}`);let r=this.temporarySessionTensorIds.get(e);if(r){for(let t of r)Dt("verbose",()=>`[WebNN] releasing temporary tensor {tensorId: ${t}}`),this.tensorManager.releaseTensorId(t);this.temporarySessionTensorIds.delete(e),this.activeSessionId=void 0}}async createMLContext(e){if(e instanceof GPUDevice){let t=this.mlContextCache.findIndex(s=>s.gpuDevice===e);if(t!==-1)return this.mlContextCache[t].mlContext;{let s=await navigator.ml.createContext(e);return this.mlContextCache.push({gpuDevice:e,mlContext:s}),s}}else if(e===void 0){let t=this.mlContextCache.findIndex(s=>s.options===void 0&&s.gpuDevice===void 0);if(t!==-1)return this.mlContextCache[t].mlContext;{let s=await navigator.ml.createContext();return this.mlContextCache.push({mlContext:s}),s}}let r=this.mlContextCache.findIndex(t=>np(t.options,e));if(r!==-1)return this.mlContextCache[r].mlContext;{let t=await navigator.ml.createContext(e);return this.mlContextCache.push({options:e,mlContext:t}),t}}registerMLContext(e,r){this.mlContextBySessionId.set(e,r);let t=this.sessionIdsByMLContext.get(r);t||(t=new Set,this.sessionIdsByMLContext.set(r,t)),t.add(e),this.temporaryGraphInputs.length>0&&(this.sessionGraphInputs.set(e,this.temporaryGraphInputs),this.temporaryGraphInputs=[])}onReleaseSession(e){this.sessionGraphInputs.delete(e);let r=this.mlContextBySessionId.get(e);if(!r)return;this.tensorManager.releaseTensorsForSession(e),this.mlContextBySessionId.delete(e);let t=this.sessionIdsByMLContext.get(r);if(t.delete(e),t.size===0){this.sessionIdsByMLContext.delete(r);let s=this.mlContextCache.findIndex(o=>o.mlContext===r);s!==-1&&this.mlContextCache.splice(s,1)}}getMLContext(e){return this.mlContextBySessionId.get(e)}reserveTensorId(){return this.tensorManager.reserveTensorId()}releaseTensorId(e){Dt("verbose",()=>`[WebNN] releaseTensorId {tensorId: ${e}}`),this.tensorManager.releaseTensorId(e)}async ensureTensor(e,r,t,s,o){let n=ua.get(t);if(!n)throw new Error(`Unsupported ONNX data type: ${t}`);return this.tensorManager.ensureTensor(e??this.currentSessionId,r,n,s,o)}async createTemporaryTensor(e,r,t){Dt("verbose",()=>`[WebNN] createTemporaryTensor {onnxDataType: ${r}, shape: ${t}}`);let s=ua.get(r);if(!s)throw new Error(`Unsupported ONNX data type: ${r}`);let o=this.tensorManager.reserveTensorId();await this.tensorManager.ensureTensor(e,o,s,t,!1);let n=this.temporarySessionTensorIds.get(e);return n?n.push(o):this.temporarySessionTensorIds.set(e,[o]),o}uploadTensor(e,r){if(!Xt().shouldTransferToMLTensor)throw new Error("Trying to upload to a MLTensor while shouldTransferToMLTensor is false");Dt("verbose",()=>`[WebNN] uploadTensor {tensorId: ${e}, data: ${r.byteLength}}`),this.tensorManager.upload(e,r)}async downloadTensor(e,r){return this.tensorManager.download(e,r)}createMLTensorDownloader(e,r){return async()=>{let t=await this.tensorManager.download(e);return rl(t,r)}}registerMLTensor(e,r,t,s){let o=ua.get(t);if(!o)throw new Error(`Unsupported ONNX data type: ${t}`);let n=this.tensorManager.registerTensor(e,r,o,s);return Dt("verbose",()=>`[WebNN] registerMLTensor {tensor: ${r}, dataType: ${o}, dimensions: ${s}} -> {tensorId: ${n}}`),n}registerMLConstant(e,r,t,s,o,n,a=!1){if(!n)throw new Error("External mounted files are not available.");let i=e;e.startsWith("./")&&(i=e.substring(2));let l=n.get(i);if(!l)throw new Error(`File with name ${i} not found in preloaded files.`);if(r+t>l.byteLength)throw new Error("Out of bounds: data offset and length exceed the external file data size.");let c=l.slice(r,r+t).buffer,p;switch(o.dataType){case"float32":p=new Float32Array(c);break;case"float16":p=typeof Float16Array<"u"&&Float16Array.from?new Float16Array(c):new Uint16Array(c);break;case"int32":p=new Int32Array(c);break;case"uint32":p=new Uint32Array(c);break;case"int64":a?(p=sl(new Uint8Array(c),!1),o.dataType="int32"):p=new BigInt64Array(c);break;case"uint64":p=new BigUint64Array(c);break;case"int8":p=new Int8Array(c);break;case"int4":case"uint4":case"uint8":p=new Uint8Array(c);break;default:throw new Error(`Unsupported data type: ${o.dataType} in creating WebNN Constant from external data.`)}return Dt("verbose",()=>`[WebNN] registerMLConstant {dataType: ${o.dataType}, shape: ${o.shape}}} ${a?"(Note: it was int64 data type and registered to int32 as workaround)":""}`),s.constant(o,p)}registerGraphInput(e){this.temporaryGraphInputs.push(e)}isGraphInput(e,r){let t=this.sessionGraphInputs.get(e);return t?t.includes(r):!1}isInt64Supported(e){var r;return!!((r=this.mlContextBySessionId.get(e))!=null&&r.opSupportLimits().input.dataTypes.includes("int64"))}flush(){}}}),ul=Ue(()=>{}),cl,ca,da,ap,ip,dl,pl,lp,up,Kx=Ue(()=>{js(),ul(),cl=new Map([[64,250],[128,200],[256,200],[512,200],[2048,230],[4096,200],[8192,50],[16384,50],[32768,50],[65536,50],[131072,50],[262144,50],[524288,50],[1048576,50],[2097152,30],[4194304,20],[8388608,10],[12582912,10],[16777216,10],[26214400,15],[33554432,22],[44236800,2],[58982400,6],[67108864,6],[134217728,6],[167772160,6]]),ca=[],da=e=>Math.ceil(Number(e)/16)*16,ap=e=>{for(let r=0;r<ca.length;r++){let t=ca[r];if(e<=t)return t}return Math.ceil(e/16)*16},ip=1,dl=()=>ip++,pl=async(e,r,t,s)=>{let o=da(t),n=e.device.createBuffer({size:o,usage:GPUBufferUsage.COPY_DST|GPUBufferUsage.MAP_READ});try{let a=e.getCommandEncoder();e.endComputePass(),a.copyBufferToBuffer(r,0,n,0,o),e.flush(),await n.mapAsync(GPUMapMode.READ);let i=n.getMappedRange();if(s){let l=s();return l.set(new Uint8Array(i,0,t)),l}else return new Uint8Array(i.slice(0,t))}finally{n.destroy()}},lp=class{constructor(e){this.backend=e,this.storageCache=new Map,this.freeBuffers=new Map,this.freeUniformBuffers=new Map,this.buffersPending=[],this.capturedPendingBuffers=new Map;for(let[r]of cl)ca.push(r),this.freeBuffers.set(r,[]),this.freeUniformBuffers.set(r,[]);this.sessionCount=0}upload(e,r){let t=r.buffer,s=r.byteOffset,o=r.byteLength,n=da(o),a=this.storageCache.get(e);if(!a)throw new Error("gpu data for uploading does not exist");if(Number(a.originalSize)!==o)throw new Error(`inconsistent data size. gpu data size=${a.originalSize}, data size=${o}`);let i=this.backend.device.createBuffer({mappedAtCreation:!0,size:n,usage:GPUBufferUsage.MAP_WRITE|GPUBufferUsage.COPY_SRC}),l=i.getMappedRange();new Uint8Array(l).set(new Uint8Array(t,s,o)),i.unmap();let c=this.backend.device.createCommandEncoder();c.copyBufferToBuffer(i,0,a.gpuData.buffer,0,n),this.backend.device.queue.submit([c.finish()]),i.destroy(),Dt("verbose",()=>`[WebGPU] GpuDataManager.upload(id=${e})`)}memcpy(e,r){let t=this.storageCache.get(e);if(!t)throw new Error("source gpu data for memcpy does not exist");let s=this.storageCache.get(r);if(!s)throw new Error("destination gpu data for memcpy does not exist");if(t.originalSize!==s.originalSize)throw new Error("inconsistent source and destination gpu data size");let o=da(t.originalSize),n=this.backend.getCommandEncoder();this.backend.endComputePass(),n.copyBufferToBuffer(t.gpuData.buffer,0,s.gpuData.buffer,0,o)}registerExternalBuffer(e,r,t){let s;if(t){if(s=t[0],e===t[1])return Dt("verbose",()=>`[WebGPU] GpuDataManager.registerExternalBuffer(size=${r}) => id=${s}, buffer is the same, skip.`),s;if(this.backend.capturedCommandList.has(this.backend.currentSessionId))throw new Error(`Registering a different external buffer under graph capture mode is not supported yet.
             Please use the previous external buffer!`)}else s=dl();return this.storageCache.set(s,{gpuData:{id:s,type:0,buffer:e},originalSize:r}),Dt("verbose",()=>`[WebGPU] GpuDataManager.registerExternalBuffer(size=${r}) => id=${s}, registered.`),s}unregisterExternalBuffer(e){e!==void 0&&(this.storageCache.delete(e),Dt("verbose",()=>`[WebGPU] GpuDataManager.unregisterExternalBuffer() => id=${e}`))}create(e,r=GPUBufferUsage.STORAGE|GPUBufferUsage.COPY_SRC|GPUBufferUsage.COPY_DST){let t=ap(e),s,o=(r&GPUBufferUsage.STORAGE)===GPUBufferUsage.STORAGE,n=(r&GPUBufferUsage.UNIFORM)===GPUBufferUsage.UNIFORM;if(o||n){let i=(o?this.freeBuffers:this.freeUniformBuffers).get(t);i?i.length>0?s=i.pop():s=this.backend.device.createBuffer({size:t,usage:r}):s=this.backend.device.createBuffer({size:t,usage:r})}else s=this.backend.device.createBuffer({size:t,usage:r});let a={id:dl(),type:0,buffer:s};return this.storageCache.set(a.id,{gpuData:a,originalSize:Number(e)}),Dt("verbose",()=>`[WebGPU] GpuDataManager.create(size=${e}) => id=${a.id}`),a}get(e){var r;return(r=this.storageCache.get(e))==null?void 0:r.gpuData}release(e){let r=typeof e=="bigint"?Number(e):e,t=this.storageCache.get(r);if(!t){if(this.storageCache.size===0)return 0;throw new Error("releasing data does not exist")}return Dt("verbose",()=>`[WebGPU] GpuDataManager.release(id=${r}), gpuDataId=${t.gpuData.id}`),this.storageCache.delete(r),this.buffersPending.push(t.gpuData.buffer),t.originalSize}async download(e,r){let t=this.storageCache.get(Number(e));if(!t)throw new Error("data does not exist");await pl(this.backend,t.gpuData.buffer,t.originalSize,r)}refreshPendingBuffers(){if(this.buffersPending.length!==0)if(this.backend.sessionStatus==="default"){for(let e of this.buffersPending){let r=cl.get(e.size);if((e.usage&GPUBufferUsage.STORAGE)===GPUBufferUsage.STORAGE){let t=this.freeBuffers.get(e.size)||[];r===void 0||t.length>=r?e.destroy():t.push(e)}else if((e.usage&GPUBufferUsage.UNIFORM)===GPUBufferUsage.UNIFORM){let t=this.freeUniformBuffers.get(e.size)||[];r===void 0||t.length>=r?e.destroy():t.push(e)}else e.destroy()}this.buffersPending=[]}else{let e=this.capturedPendingBuffers.get(this.backend.currentSessionId);e||(e=[],this.capturedPendingBuffers.set(this.backend.currentSessionId,e));for(let r of this.buffersPending)e.push(r);this.buffersPending=[]}}dispose(){this.freeBuffers.forEach(e=>{e.forEach(r=>{r.destroy()})}),this.freeUniformBuffers.forEach(e=>{e.forEach(r=>{r.destroy()})}),this.storageCache.forEach(e=>{e.gpuData.buffer.destroy()}),this.capturedPendingBuffers.forEach(e=>{e.forEach(r=>{r.destroy()})}),this.storageCache=new Map,this.freeBuffers=new Map,this.freeUniformBuffers=new Map,this.capturedPendingBuffers=new Map}onCreateSession(){this.sessionCount+=1}onReleaseSession(e){let r=this.capturedPendingBuffers.get(e);r&&(r.forEach(t=>{t.destroy()}),this.capturedPendingBuffers.delete(e)),this.sessionCount-=1,this.sessionCount===0&&(Dt("warning",()=>"[WebGPU] Clearing webgpu buffer cache"),this.storageCache.forEach(t=>{t.gpuData.buffer.destroy()}),this.storageCache=new Map)}},up=(...e)=>new lp(...e)}),cp,jt,ur=Ue(()=>{cp=class{constructor(e){Object.assign(this,e)}get cacheKey(){return this.key||(this.key=Object.getOwnPropertyNames(this).sort().map(e=>`${this[e]}`).join(";")),this.key}},jt=e=>new cp(e)}),Hn,pa,Sr,jr,lt,or,hl,Kn,Js,it,go,Oe,at,dp,ml,pp,hp,Ct=Ue(()=>{Mt(),Pt(),Hn=64,pa=(e,r)=>{if(r===3)throw new Error("vec3 has same alignment as vec4, use vec4 instead");switch(Number(e)){case 10:return r>1?`vec${r}<f16>`:"f16";case 1:return r>1?`vec${r}<f32>`:"f32";case 6:return r>1?`vec${r}<i32>`:"i32";case 12:return r>1?`vec${r}<u32>`:"u32";case 7:if(r>1)throw new Error("currently not supported vecX of uint64 yet");return["vec2<u32>","i32"];case 13:if(r>1)throw new Error("currently not supported vecX of uint64 yet");return["vec2<u32>","u32"];case 9:if(r!==4)throw new Error("bool must be vec4");return["u32","vec4<bool>"];case 22:return"i32";case 21:return"u32";default:throw new Error(`Unknown data type: ${e}`)}},Sr=(e,r=1)=>{let t=pa(e,r);return typeof t=="string"?t:t[0]},jr=(e,r=1)=>{let t=pa(e,r);return typeof t=="string"?t:t[1]},lt=(...e)=>{let r=[];return e.forEach(t=>{t.length!==0&&r.push({type:12,data:t},{type:12,data:Pe.computeStrides(t)})}),r},or=e=>e%4===0?4:e%2===0?2:1,hl=(e="f32",r,t="0")=>!r||r===1?`${e}(${t})`:`vec${r}<${e}>(${t})`,Kn=(e,r,t)=>e==="f32"?t:r===1?`f32(${t})`:`vec${r}<f32>(${t})`,Js=(e,r)=>r===4?`(${e}.x + ${e}.y + ${e}.z + ${e}.w)`:r===2?`(${e}.x + ${e}.y)`:r===3?`(${e}.x + ${e}.y + ${e}.z)`:e,it=(e,r,t,s)=>e.startsWith("uniforms.")&&t>4?typeof r=="string"?s==="f16"?`${e}[(${r}) / 8][(${r}) % 8 / 4][(${r}) % 8 % 4]`:`${e}[(${r}) / 4][(${r}) % 4]`:s==="f16"?`${e}[${Math.floor(r/8)}][${Math.floor(r%8/4)}][${r%8%4}]`:`${e}[${Math.floor(r/4)}][${r%4}]`:t>1?`${e}[${r}]`:e,go=(e,r,t,s,o)=>{let n=typeof t=="number",a=n?t:t.length,i=[...new Array(a).keys()],l=a<2?"u32":a<=4?`vec${a}<u32>`:`array<u32, ${a}>`,c=pa(r,o),p=typeof c=="string"?c:c[1],d=typeof c=="string"?c:c[0],u={indices:l,value:p,storage:d,tensor:r},_=B=>typeof B=="string"?B:`${B}u`,f={offsetToIndices:!1,indicesToOffset:!1,broadcastedIndicesToOffset:!1,set:!1,setByIndices:!1,get:!1,getByIndices:!1},b=n?"uniforms.":"",A=`${b}${e}_shape`,g=`${b}${e}_strides`,y="";for(let B=0;B<a-1;B++)y+=`
    let dim${B} = current / ${it(g,B,a)};
    let rest${B} = current % ${it(g,B,a)};
    indices[${B}] = dim${B};
    current = rest${B};
    `;y+=`indices[${a-1}] = current;`;let C=a<2?"":`
  fn o2i_${e}(offset: u32) -> ${u.indices} {
    var indices: ${u.indices};
    var current = offset;
    ${y}
    return indices;
  }`,x=B=>(f.offsetToIndices=!0,a<2?B:`o2i_${e}(${B})`),M=[];if(a>=2)for(let B=a-1;B>=0;B--)M.push(`${it(g,B,a)} * (indices[${B}])`);let T=a<2?"":`
  fn i2o_${e}(indices: ${u.indices}) -> u32 {
    return ${M.join("+")};
  }`,v=B=>(f.indicesToOffset=!0,a<2?B:`i2o_${e}(${B})`),P=(...B)=>a===0?"0u":`${u.indices}(${B.map(_).join(",")})`,F=(B,O)=>a<2?`${B}`:`${it(B,O,a)}`,D=(B,O,W)=>a<2?`${B}=${W};`:`${it(B,O,a)}=${W};`,K={},U=(B,O)=>{f.broadcastedIndicesToOffset=!0;let W=`${O.name}broadcastedIndicesTo${e}Offset`;if(W in K)return`${W}(${B})`;let N=[];for(let J=a-1;J>=0;J--){let ie=O.indicesGet("outputIndices",J+O.rank-a);N.push(`${F(g,J)} * (${ie} % ${F(A,J)})`)}return K[W]=`fn ${W}(outputIndices: ${O.type.indices}) -> u32 {
             return ${N.length>0?N.join("+"):"0u"};
           }`,`${W}(${B})`},j=(B,O)=>(()=>{if(u.storage===u.value)return`${e}[${B}]=${O};`;if(u.storage==="vec2<u32>"&&u.value==="i32")return`${e}[${B}]=vec2<u32>(u32(${O}), select(0u, 0xFFFFFFFFu, ${O} < 0));`;if(u.storage==="vec2<u32>"&&u.value==="u32")return`${e}[${B}]=vec2<u32>(u32(${O}), 0u);`;if(u.storage==="u32"&&u.value==="vec4<bool>")return`${e}[${B}]=dot(vec4<u32>(0x1, 0x100, 0x10000, 0x1000000), vec4<u32>(${O}));`;throw new Error(`not supported combination of storage type ${u.storage} and value type ${u.value} yet`)})(),ne=B=>(()=>{if(u.storage===u.value)return`${e}[${B}]`;if(u.storage==="vec2<u32>"&&u.value==="i32")return`i32(${e}[${B}].x)`;if(u.storage==="vec2<u32>"&&u.value==="u32")return`u32(${e}[${B}].x)`;if(u.storage==="u32"&&u.value==="vec4<bool>")return`vec4<bool>(bool(${e}[${B}] & 0xFFu), bool(${e}[${B}] & 0xFF00u), bool(${e}[${B}] & 0xFF0000u), bool(${e}[${B}] & 0xFF000000u))`;throw new Error(`not supported combination of storage type ${u.storage} and value type ${u.value} yet`)})(),q=a<2?"":`
  fn get_${e}ByIndices(indices: ${u.indices}) -> ${p} {
    return ${ne(`i2o_${e}(indices)`)};
  }`,te=a<2?"":(()=>{let B=i.map(W=>`d${W}: u32`).join(", "),O=i.map(W=>`d${W}`).join(", ");return`
  fn get_${e}(${B}) -> ${p} {
    return get_${e}ByIndices(${P(O)});
  }`})(),Z=(...B)=>{if(B.length!==a)throw new Error(`indices length must be ${a}`);let O=B.map(_).join(",");return a===0?ne("0u"):a===1?ne(O[0]):(f.get=!0,f.getByIndices=!0,f.indicesToOffset=!0,`get_${e}(${O})`)},ae=B=>a<2?ne(B):(f.getByIndices=!0,f.indicesToOffset=!0,`get_${e}ByIndices(${B})`),he=a<2?"":`
  fn set_${e}ByIndices(indices: ${u.indices}, value: ${p}) {
    ${j(`i2o_${e}(indices)`,"value")}
  }`,Q=a<2?"":(()=>{let B=i.map(W=>`d${W}: u32`).join(", "),O=i.map(W=>`d${W}`).join(", ");return`
  fn set_${e}(${B}, value: ${p}) {
    set_${e}ByIndices(${P(O)}, value);
  }`})();return{impl:()=>{let B=[],O=!1;return f.offsetToIndices&&(B.push(C),O=!0),f.indicesToOffset&&(B.push(T),O=!0),f.broadcastedIndicesToOffset&&(Object.values(K).forEach(W=>B.push(W)),O=!0),f.set&&(B.push(Q),O=!0),f.setByIndices&&(B.push(he),O=!0),f.get&&(B.push(te),O=!0),f.getByIndices&&(B.push(q),O=!0),!n&&O&&B.unshift(`const ${A} = ${u.indices}(${t.join(",")});`,`const ${g} = ${u.indices}(${Pe.computeStrides(t).join(",")});`),B.join(`
`)},type:u,offsetToIndices:x,indicesToOffset:v,broadcastedIndicesToOffset:U,indices:P,indicesGet:F,indicesSet:D,set:(...B)=>{if(B.length!==a+1)throw new Error(`indices length must be ${a}`);let O=B[a];if(typeof O!="string")throw new Error("value must be string");let W=B.slice(0,a).map(_).join(",");return a===0?j("0u",O):a===1?j(W[0],O):(f.set=!0,f.setByIndices=!0,f.indicesToOffset=!0,`set_${e}(${W}, ${O})`)},setByOffset:j,setByIndices:(B,O)=>a<2?j(B,O):(f.setByIndices=!0,f.indicesToOffset=!0,`set_${e}ByIndices(${B}, ${O});`),get:Z,getByOffset:ne,getByIndices:ae,usage:s,name:e,strides:g,shape:A,rank:a}},Oe=(e,r,t,s=1)=>go(e,r,t,"input",s),at=(e,r,t,s=1)=>go(e,r,t,"output",s),dp=(e,r,t)=>go(e,r,t,"atomicOutput",1),ml=(e,r,t,s=1)=>go(e,r,t,"internal",s),pp=class{constructor(e,r){this.normalizedDispatchGroup=e,this.limits=r,this.internalVariables=[],this.variables=[],this.uniforms=[],this.variableIndex=0}guardAgainstOutOfBoundsWorkgroupSizes(e){return`if (global_idx >= ${typeof e=="number"?`${e}u`:e}) { return; }`}mainStart(e=Hn){let r=typeof e=="number"?e:e[0],t=typeof e=="number"?1:e[1],s=typeof e=="number"?1:e[2];if(r>this.limits.maxComputeWorkgroupSizeX||t>this.limits.maxComputeWorkgroupSizeY||s>this.limits.maxComputeWorkgroupSizeZ)throw new Error(`workgroup size [${r}, ${t}, ${s}] exceeds the maximum workgroup size [${this.limits.maxComputeWorkgroupSizeX}, ${this.limits.maxComputeWorkgroupSizeY}, ${this.limits.maxComputeWorkgroupSizeZ}].`);if(r*t*s>this.limits.maxComputeInvocationsPerWorkgroup)throw new Error(`workgroup size [${r}, ${t}, ${s}] exceeds the maximum workgroup invocations ${this.limits.maxComputeInvocationsPerWorkgroup}.`);let o=this.normalizedDispatchGroup[1]===1&&this.normalizedDispatchGroup[2]===1,n=o?`@builtin(global_invocation_id) global_id : vec3<u32>,
    @builtin(workgroup_id) workgroup_id : vec3<u32>,
    @builtin(local_invocation_index) local_idx : u32,
    @builtin(local_invocation_id) local_id : vec3<u32>`:`@builtin(global_invocation_id) global_id : vec3<u32>,
                                             @builtin(local_invocation_id) local_id : vec3<u32>,
    @builtin(local_invocation_index) local_idx : u32,
    @builtin(workgroup_id) workgroup_id : vec3<u32>,
    @builtin(num_workgroups) num_workgroups : vec3<u32>`,a=o?`let global_idx = global_id.x;
         let workgroup_index = workgroup_id.x;`:`let workgroup_index = workgroup_id.z * num_workgroups[0] * num_workgroups[1] +
             workgroup_id.y * num_workgroups[0] + workgroup_id.x;
         let global_idx = workgroup_index * ${r*t*s}u + local_idx;`;return`@compute @workgroup_size(${r}, ${t}, ${s})
  fn main(${n}) {
    ${a}
  `}appendVariableUniforms(e){e.rank!==0&&(e.shape.startsWith("uniforms.")&&this.uniforms.push({name:e.shape.replace("uniforms.",""),type:"u32",length:e.rank}),e.strides.startsWith("uniforms.")&&this.uniforms.push({name:e.strides.replace("uniforms.",""),type:"u32",length:e.rank}))}declareVariable(e,r){if(e.usage==="internal")throw new Error("cannot use internal variable with declareVariable(). use registerInternalVariables() instead.");this.variables.push(e),this.appendVariableUniforms(e);let t=e.usage==="input"?"read":"read_write",s=e.usage==="atomicOutput"?"atomic<i32>":e.type.storage;return`@group(0) @binding(${r}) var<storage, ${t}> ${e.name}: array<${s}>;`}declareVariables(...e){return e.map(r=>this.declareVariable(r,this.variableIndex++)).join(`
`)}registerInternalVariable(e){if(e.usage!=="internal")throw new Error("cannot use input or output variable with registerInternalVariable(). use declareVariables() instead.");this.internalVariables.push(e),this.appendVariableUniforms(e)}registerInternalVariables(...e){return e.forEach(r=>this.registerInternalVariable(r)),this}registerUniform(e,r,t=1){return this.uniforms.push({name:e,type:r,length:t}),this}registerUniforms(e){return this.uniforms=this.uniforms.concat(e),this}uniformDeclaration(){if(this.uniforms.length===0)return"";let e=[];for(let{name:r,type:t,length:s}of this.uniforms)if(s&&s>4)t==="f16"?e.push(`@align(16) ${r}:array<mat2x4<${t}>, ${Math.ceil(s/8)}>`):e.push(`${r}:array<vec4<${t}>, ${Math.ceil(s/4)}>`);else{let o=s==null||s===1?t:`vec${s}<${t}>`;e.push(`${r}:${o}`)}return`
      struct Uniforms { ${e.join(", ")} };
      @group(0) @binding(${this.variableIndex}) var<uniform> uniforms: Uniforms;`}get additionalImplementations(){return this.uniformDeclaration()+this.variables.map(e=>e.impl()).join(`
`)+this.internalVariables.map(e=>e.impl()).join(`
`)}get variablesInfo(){if(this.uniforms.length===0)return;let e=r=>[12,10,1,6][["u32","f16","f32","i32"].indexOf(r)];return this.uniforms.map(r=>[e(r.type),r.length??1])}},hp=(e,r)=>new pp(e,r)}),mp,fl,fp,_p,gp,wp,rs,Mp,bp,Ys=Ue(()=>{Mt(),Pt(),ur(),Ct(),mp=(e,r)=>{if(!e||e.length!==1)throw new Error("Transpose requires 1 input.");if(r.length!==0&&r.length!==e[0].dims.length)throw new Error(`perm size ${r.length} does not match input rank ${e[0].dims.length}`)},fl=(e,r)=>r.length!==0?r:[...new Array(e).keys()].reverse(),fp=(e,r)=>Pe.sortBasedOnPerm(e,fl(e.length,r)),_p=(e,r,t,s)=>{let o=`fn perm(i: ${s.type.indices}) -> ${t.type.indices} {
    var a: ${t.type.indices};`;for(let n=0;n<r;++n)o+=`a[${e[n]}]=i[${n}];`;return o+="return a;}"},gp=(e,r)=>{let t=[],s=[];for(let o=0;o<e.length;++o)e[o]!==1&&t.push(e[o]),e[r[o]]!==1&&s.push(r[o]);return{newShape:t,newPerm:s}},wp=(e,r)=>{let t=0;for(let s=0;s<e.length;++s)if(r[e[s]]!==1){if(e[s]<t)return!1;t=e[s]}return!0},rs=(e,r)=>{let t=e.dataType,s=e.dims.length,o=fl(s,r),n=fp(e.dims,o),a=e.dims,i=n,l=s<2||wp(o,e.dims),c;if(l)return c=f=>{let b=Oe("input",t,a,4),A=at("output",t,i,4);return`
  ${f.registerUniform("output_size","u32").declareVariables(b,A)}
  ${f.mainStart()}
    ${f.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_size")}
    output[global_idx] = input[global_idx];
  }`},{name:"TransposeCopy",shaderCache:{inputDependencies:["type"]},getRunData:()=>{let f=Pe.size(n);return{outputs:[{dims:n,dataType:e.dataType}],dispatchGroup:{x:Math.ceil(f/64/4)},programUniforms:[{type:12,data:Math.ceil(f/4)}]}},getShaderSource:c};let{newShape:p,newPerm:d}=gp(e.dims,o),u=Pe.areEqual(d,[2,3,1]),_=Pe.areEqual(d,[3,1,2]);if(p.length===2||u||_){a=u?[p[0],p[1]*p[2]]:_?[p[0]*p[1],p[2]]:p,i=[a[1],a[0]];let f=16;return c=b=>{let A=Oe("a",t,a.length),g=at("output",t,i.length);return`
  ${b.registerUniform("output_size","u32").declareVariables(A,g)}
  var<workgroup> tile : array<array<${g.type.value}, ${f+1}>, ${f}>;
  ${b.mainStart([f,f,1])}
    let stride = (uniforms.output_shape[1] - 1) / ${f} + 1;
    let workgroup_id_x = workgroup_index % stride;
    let workgroup_id_y = workgroup_index / stride;
    let input_col = workgroup_id_y * ${f}u + local_id.x;
    let input_row = workgroup_id_x * ${f}u + local_id.y;
    if (input_row < uniforms.a_shape[0] && input_col < uniforms.a_shape[1]) {
      tile[local_id.y][local_id.x] = ${A.getByIndices(`${A.type.indices}(input_row, input_col)`)};
    }
    workgroupBarrier();

    let output_col = workgroup_id_x * ${f}u + local_id.x;
    let output_row = workgroup_id_y * ${f}u + local_id.y;
    if (output_row < uniforms.output_shape[0] && output_col < uniforms.output_shape[1]) {
      ${g.setByIndices(`${g.type.indices}(output_row, output_col)`,"tile[local_id.x][local_id.y]")}
    }
  }`},{name:"TransposeShared",shaderCache:{inputDependencies:["type"]},getRunData:()=>{let b=Pe.size(n);return{outputs:[{dims:n,dataType:e.dataType}],dispatchGroup:{x:Math.ceil(i[1]/f),y:Math.ceil(i[0]/f)},programUniforms:[{type:12,data:b},...lt(a,i)]}},getShaderSource:c}}return c=f=>{let b=Oe("a",t,a.length),A=at("output",t,i.length);return`
  ${f.registerUniform("output_size","u32").declareVariables(b,A)}

  ${_p(o,s,b,A)}

  ${f.mainStart()}
    ${f.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_size")}

    let indices = ${A.offsetToIndices("global_idx")};
    let aIndices = perm(indices);

    ${A.setByOffset("global_idx",b.getByIndices("aIndices"))}
  }`},{name:"Transpose",shaderCache:{hint:`${r}`,inputDependencies:["rank"]},getRunData:()=>{let f=Pe.size(n);return{outputs:[{dims:n,dataType:e.dataType}],dispatchGroup:{x:Math.ceil(f/64)},programUniforms:[{type:12,data:f},...lt(a,i)]}},getShaderSource:c}},Mp=(e,r)=>{mp(e.inputs,r.perm),e.compute(rs(e.inputs[0],r.perm))},bp=e=>jt({perm:e.perm})}),yp,vp,xp,Tp,Ep,Pp,Cp,Sp,$p,Ap,vs,kp,Ip,Fp,Op,Dp,Lp,zp,Rp,Bp,Np,qx=Ue(()=>{Mt(),Pt(),Ct(),gl(),Ys(),yp={max:"select(bestValue, candidate, candidate > bestValue)",min:"select(bestValue, candidate, candidate < bestValue)",mean:"bestValue + candidate",sum:"bestValue + candidate",prod:"bestValue * candidate",sumSquare:"bestValue + candidate * candidate",logSumExp:"bestValue + exp(candidate)",l1:"bestValue + abs(candidate)",l2:"bestValue + candidate * candidate",logSum:"bestValue + candidate"},vp={max:"select(bestValue, candidate, candidate > bestValue)",min:"select(bestValue, candidate, candidate < bestValue)",mean:"bestValue + candidate",sum:"bestValue + candidate",prod:"bestValue * candidate",sumSquare:"bestValue + candidate",logSumExp:"bestValue + candidate",l1:"bestValue + candidate",l2:"bestValue + candidate",logSum:"bestValue + candidate"},xp={max:"_A[offset]",min:"_A[offset]",mean:"0",sum:"0",prod:"1",sumSquare:"0",logSumExp:"0",l1:"0",l2:"0",logSum:"0"},Tp={max:"bestValue",min:"bestValue",sum:"bestValue",prod:"bestValue",sumSquare:"bestValue",logSumExp:"log(bestValue)",l1:"bestValue",l2:"sqrt(bestValue)",logSum:"log(bestValue)"},Ep=(e,r)=>{let t=[];for(let s=r-e;s<r;++s)t.push(s);return t},Pp=(e,r)=>{let t=[],s=e.length;for(let n=0;n<s;n++)r.indexOf(n)===-1&&t.push(e[n]);let o=r.map(n=>e[n]);return[t,o]},Cp=(e,r)=>{let t=e.length+r.length,s=[],o=0;for(let n=0;n<t;n++)r.indexOf(n)===-1?s.push(e[o++]):s.push(1);return s},Sp=(e,r)=>{for(let t=0;t<e.length;++t)if(e[e.length-t-1]!==r-1-t)return!1;return!0},$p=(e,r)=>{let t=[];if(!Sp(e,r)){for(let s=0;s<r;++s)e.indexOf(s)===-1&&t.push(s);e.forEach(s=>t.push(s))}return t},Ap=(e,r,t,s,o,n,a)=>{let i=t[0].dims,l=Pe.size(n),c=Pe.size(a),p=Oe("_A",t[0].dataType,i),d=at("output",o,n),u=64;l===1&&(u=256);let _=`
          var<workgroup> aBestValues : array<f32, ${u}>;
       `,f=b=>`
        ${b.registerUniform("reduceSize","u32").declareVariables(p,d)}
        ${_}
        fn DIV_CEIL(a : u32, b : u32) -> u32 {
          return ((a - 1u) / b + 1u);
         }
         ${b.mainStart(u)}

          let outputIndex = global_idx / ${u};
          let offset = outputIndex * uniforms.reduceSize;

          var bestValue = f32(${xp[s]});
          let Length = uniforms.reduceSize;
          for (var k = local_idx; k < Length; k = k + ${u}) {
           let candidate = f32(${p.getByOffset("offset + k")});
           bestValue = ${yp[s]};
          }
          aBestValues[local_idx] = bestValue;
          workgroupBarrier();

         var reduceSize = min(Length, ${u}u);
         for (var currentSize = reduceSize / 2u; reduceSize > 1u;
             currentSize = reduceSize / 2u) {
           let interval = DIV_CEIL(reduceSize, 2u);
           if (local_idx < currentSize) {
            let candidate = aBestValues[local_idx + interval];
            bestValue = ${vp[s]};
            aBestValues[local_idx] = bestValue;
           }
           reduceSize = interval;
           workgroupBarrier();
         }

         if (local_idx == 0u) {
          ${d.setByOffset("outputIndex",`${s==="mean"?`${d.type.storage}(bestValue / f32(uniforms.reduceSize))`:`${d.type.storage}(${Tp[s]})`}`)};
         }
        }`;return{name:e,shaderCache:{hint:`${r};${u}`,inputDependencies:["type"]},getShaderSource:f,getRunData:()=>({outputs:[{dims:n,dataType:o}],dispatchGroup:{x:l},programUniforms:[{type:12,data:c}]})}},vs=(e,r,t,s)=>{let o=e.inputs.length===1?t:_l(e.inputs,t),n=o.axes;n.length===0&&!o.noopWithEmptyAxes&&(n=e.inputs[0].dims.map((_,f)=>f));let a=Pe.normalizeAxes(n,e.inputs[0].dims.length),i=a,l=e.inputs[0],c=$p(i,e.inputs[0].dims.length);c.length>0&&(l=e.compute(rs(e.inputs[0],c),{inputs:[0],outputs:[-1]})[0],i=Ep(i.length,l.dims.length));let[p,d]=Pp(l.dims,i),u=p;o.keepDims&&(u=Cp(p,a)),e.compute(Ap(r,o.cacheKey,[l],s,e.inputs[0].dataType,u,d),{inputs:[l]})},kp=(e,r)=>{vs(e,"ReduceMeanShared",r,"mean")},Ip=(e,r)=>{vs(e,"ReduceL1Shared",r,"l1")},Fp=(e,r)=>{vs(e,"ReduceL2Shared",r,"l2")},Op=(e,r)=>{vs(e,"ReduceLogSumExpShared",r,"logSumExp")},Dp=(e,r)=>{vs(e,"ReduceMaxShared",r,"max")},Lp=(e,r)=>{vs(e,"ReduceMinShared",r,"min")},zp=(e,r)=>{vs(e,"ReduceProdShared",r,"prod")},Rp=(e,r)=>{vs(e,"ReduceSumShared",r,"sum")},Bp=(e,r)=>{vs(e,"ReduceSumSquareShared",r,"sumSquare")},Np=(e,r)=>{vs(e,"ReduceLogSumShared",r,"logSum")}}),xs,jp,ha,_l,Ts,Vp,Up,Wp,Gp,Hp,Kp,qp,Qp,Xp,Jp,Es,Yp,Zp,eh,th,rh,sh,nh,oh,ah,ih,gl=Ue(()=>{Mt(),Pt(),ur(),Ct(),qx(),xs=e=>{if(!e||e.length===0||e.length>2)throw new Error("Reduce op requires 1 or 2 inputs.");if(e.length===2&&e[1].dims.length!==1)throw new Error("Invalid axes input dims.")},jp=e=>["","",`var value = ${e.getByIndices("input_indices")};`,""],ha=(e,r,t,s,o,n,a=!1,i=!1)=>{let l=[],c=t[0].dims,p=c.length,d=Pe.normalizeAxes(o,p),u=!i&&d.length===0;c.forEach((b,A)=>{u||d.indexOf(A)>=0?a&&l.push(1):l.push(b)});let _=l.length,f=Pe.size(l);return{name:e,shaderCache:r,getShaderSource:b=>{let A=[],g=Oe("_A",t[0].dataType,p),y=at("output",n,_),C=s(g,y,d),x=C[2];for(let M=0,T=0;M<p;M++)u||d.indexOf(M)>=0?(a&&T++,x=`for(var j${M}: u32 = 0; j${M} < ${c[M]}; j${M}++) {
                  ${C[2].includes("last_index")?`let last_index = j${M};`:""}
                  ${g.indicesSet("input_indices",M,`j${M}`)}
                  ${x}
                }`):(A.push(`${g.indicesSet("input_indices",M,y.indicesGet("output_indices",T))};`),T++);return`

        ${b.registerUniform("output_size","u32").declareVariables(g,y)}

        ${b.mainStart()}
          ${b.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_size")}
          var input_indices: ${g.type.indices};
          let output_indices = ${y.offsetToIndices("global_idx")};

          ${A.join(`
`)}
          ${C[0]}       // init ops for reduce max/min
          ${C[1]}
          ${x}
          ${C[3]}
          ${C.length===4?y.setByOffset("global_idx","value"):C.slice(4).join(`
`)}
        }`},getRunData:()=>({outputs:[{dims:l,dataType:n}],dispatchGroup:{x:Math.ceil(f/64)},programUniforms:[{type:12,data:f},...lt(c,l)]})}},_l=(e,r)=>{let t=[];return e[1].dims[0]>0&&e[1].getBigInt64Array().forEach(s=>t.push(Number(s))),jt({axes:t,keepDims:r.keepDims,noopWithEmptyAxes:r.noopWithEmptyAxes})},Ts=(e,r,t,s)=>{let o=e.inputs,n=o.length===1?t:_l(o,t);e.compute(ha(r,{hint:n.cacheKey,inputDependencies:["rank"]},[o[0]],n.noopWithEmptyAxes&&n.axes.length===0?jp:s,n.axes,o[0].dataType,n.keepDims,n.noopWithEmptyAxes),{inputs:[0]})},Vp=(e,r)=>{xs(e.inputs),Ts(e,"ReduceLogSum",r,(t,s)=>[`var value = ${s.type.storage}(0);`,"",`value += ${t.getByIndices("input_indices")};`,"value = log(value);"])},Up=(e,r)=>{xs(e.inputs),Ts(e,"ReduceL1",r,(t,s)=>[`var value = ${s.type.storage}(0);`,"",`value += abs(${t.getByIndices("input_indices")});`,""])},Wp=(e,r)=>{xs(e.inputs),Ts(e,"ReduceL2",r,(t,s)=>[`var t = ${s.type.value}(0); var value = ${s.type.value}(0);`,"",`t = ${t.getByIndices("input_indices")}; value += (t * t);`,"value = sqrt(value);"])},Gp=(e,r)=>{xs(e.inputs),Ts(e,"ReduceLogSumExp",r,(t,s)=>[`var value = ${s.type.storage}(0);`,"",`value += exp(${t.getByIndices("input_indices")});`,"value = log(value);"])},Hp=(e,r)=>{xs(e.inputs),Ts(e,"ReduceMax",r,(t,s,o)=>{let n=[];for(let a=0;a<t.rank;a++)(o.indexOf(a)>=0||o.length===0)&&n.push(t.indicesSet("input_indices",a,0));return[`${n.join(`
`)}`,`var value = ${t.getByIndices("input_indices")};`,`value = max(value, ${t.getByIndices("input_indices")});`,""]})},Kp=(e,r)=>{xs(e.inputs),Ts(e,"ReduceMean",r,(t,s,o)=>{let n=1;for(let a=0;a<t.rank;a++)(o.indexOf(a)>=0||o.length===0)&&(n*=e.inputs[0].dims[a]);return["var sum = f32(0);","",`sum += f32(${t.getByIndices("input_indices")});`,`let value = ${s.type.value}(sum / ${n});`]})},qp=(e,r)=>{xs(e.inputs),Ts(e,"ReduceMin",r,(t,s,o)=>{let n=[];for(let a=0;a<t.rank;a++)(o.indexOf(a)>=0||o.length===0)&&n.push(`input_indices[${a}] = 0;`);return[`${n.join(`
`)}`,`var value = ${t.getByIndices("input_indices")};`,`value = min(value, ${t.getByIndices("input_indices")});`,""]})},Qp=(e,r)=>{xs(e.inputs),Ts(e,"ReduceProd",r,(t,s)=>[`var value = ${s.type.storage}(1);`,"",`value *= ${t.getByIndices("input_indices")};`,""])},Xp=(e,r)=>{xs(e.inputs),Ts(e,"ReduceSum",r,(t,s)=>[`var value = ${s.type.storage}(0);`,"",`value += ${t.getByIndices("input_indices")};`,""])},Jp=(e,r)=>{xs(e.inputs),Ts(e,"ReduceSumSquare",r,(t,s)=>[`var t = ${s.type.value}(0); var value = ${s.type.value}(0);`,"",`t = ${t.getByIndices("input_indices")}; value += t * t;`,""])},Es=(e,r,t)=>{if(r.length===0)return t;let s=1,o=1;for(let n=0;n<r.length;n++)r.indexOf(n)===-1?s*=e[n]:o*=e[n];return o<32&&s>1024},Yp=(e,r)=>{Es(e.inputs[0].dims,r.axes,r.noopWithEmptyAxes)?Kp(e,r):kp(e,r)},Zp=(e,r)=>{Es(e.inputs[0].dims,r.axes,r.noopWithEmptyAxes)?Up(e,r):Ip(e,r)},eh=(e,r)=>{Es(e.inputs[0].dims,r.axes,r.noopWithEmptyAxes)?Wp(e,r):Fp(e,r)},th=(e,r)=>{Es(e.inputs[0].dims,r.axes,r.noopWithEmptyAxes)?Gp(e,r):Op(e,r)},rh=(e,r)=>{Es(e.inputs[0].dims,r.axes,r.noopWithEmptyAxes)?Hp(e,r):Dp(e,r)},sh=(e,r)=>{Es(e.inputs[0].dims,r.axes,r.noopWithEmptyAxes)?qp(e,r):Lp(e,r)},nh=(e,r)=>{Es(e.inputs[0].dims,r.axes,r.noopWithEmptyAxes)?Qp(e,r):zp(e,r)},oh=(e,r)=>{Es(e.inputs[0].dims,r.axes,r.noopWithEmptyAxes)?Xp(e,r):Rp(e,r)},ah=(e,r)=>{Es(e.inputs[0].dims,r.axes,r.noopWithEmptyAxes)?Jp(e,r):Bp(e,r)},ih=(e,r)=>{Es(e.inputs[0].dims,r.axes,r.noopWithEmptyAxes)?Vp(e,r):Np(e,r)}}),wl,lh,uh,Ml,Qx=Ue(()=>{Mt(),ur(),gl(),wl=e=>{if(!e||e.length===0||e.length>2)throw new Error("ArgMinMaxOp op requires 1 or 2 inputs.");if(e[0].dataType!==1)throw new Error("Invalid input type.")},lh=(e,r)=>{wl(e.inputs);let t=(s,o,n)=>{let a=[];for(let i=0;i<s.rank;i++)(n.indexOf(i)>=0||n.length===0)&&a.push(`input_indices[${i}] = 0;`);return[`${a.join(`
`)}`,`var value = ${s.getByIndices("input_indices")};
var best_index : i32 = 0;`,`if (${s.getByIndices("input_indices")} ${r.selectLastIndex>0?"<=":"<"} value) {
         value = ${s.getByIndices("input_indices")};
         best_index = i32(last_index);
       }`,"",o.setByOffset("global_idx","best_index")]};e.compute(ha("ArgMin",{hint:r.cacheKey,inputDependencies:["rank"]},[e.inputs[0]],t,[r.axis],7,r.keepDims),{inputs:[0]})},uh=(e,r)=>{wl(e.inputs);let t=(s,o,n)=>{let a=[];for(let i=0;i<s.rank;i++)(n.indexOf(i)>=0||n.length===0)&&a.push(`input_indices[${i}] = 0;`);return[`${a.join(`
`)}`,`var value = ${s.getByIndices("input_indices")};
var best_index : i32 = 0;`,`if (${s.getByIndices("input_indices")} ${r.selectLastIndex>0?">=":">"} value) {
         value = ${s.getByIndices("input_indices")};
         best_index = i32(last_index);
       }`,"",o.setByOffset("global_idx","best_index")]};e.compute(ha("argMax",{hint:r.cacheKey,inputDependencies:["rank"]},[e.inputs[0]],t,[r.axis],7,r.keepDims),{inputs:[0]})},Ml=e=>jt(e)}),ch,ma,dh,ph,hh,wo,mh,fh,bl=Ue(()=>{Mt(),Pt(),ul(),Ct(),ch=(e,r)=>{let t=e[0],s=e[1],o=e[2],n=e[3],a=e[4],i=e[5];if(a&&i)throw new Error("Attention cannot have both past and attention_bias");if(t.dims.length!==3)throw new Error('Input "input" must have 3 dimensions');let l=t.dims[0],c=t.dims[1],p=t.dims[2];if(o.dims.length!==1)throw new Error('Input "bias" is expected to have 1 dimensions');if(s.dims.length!==2)throw new Error('Input "weights" is expected to have 2 dimensions');if(s.dims[0]!==p)throw new Error("Input 1 dimension 0 should have same length as dimension 2 of input 0");if(o.dims[0]!==s.dims[1])throw new Error('Input "bias" dimension 0 should have same length as dimension 1 of input "weights"');let d=o.dims[0]/3,u=d,_=u;if(r.qkvHiddenSizes.length>0){if(r.qkvHiddenSizes.length!==3)throw new Error("qkv_hidden_sizes attribute should have 3 elements");for(let C of r.qkvHiddenSizes)if(C%r.numHeads!==0)throw new Error("qkv_hidden_sizes should be divisible by num_heads");d=r.qkvHiddenSizes[0],u=r.qkvHiddenSizes[1],_=r.qkvHiddenSizes[2]}let f=c;if(d!==u)throw new Error("qkv_hidden_sizes first element should be same as the second");if(o.dims[0]!==d+u+_)throw new Error('Input "bias" dimension 0 should have same length as sum of Q/K/V hidden sizes');let b=0;if(a){if(u!==_)throw new Error('Input "past" expect k_hidden_size == v_hidden_size');if(a.dims.length!==5)throw new Error('Input "past" must have 5 dimensions');if(a.dims[0]!==2)throw new Error('Input "past" first dimension must be 2');if(a.dims[1]!==l)throw new Error('Input "past" second dimension must be batch_size');if(a.dims[2]!==r.numHeads)throw new Error('Input "past" third dimension must be num_heads');if(a.dims[4]!==u/r.numHeads)throw new Error('Input "past" fifth dimension must be k_hidden_size / num_heads');r.pastPresentShareBuffer||(b=a.dims[3])}let A=f+b,g=-1,y=0;if(n)throw new Error("Mask not supported");if(a)throw new Error("past is not supported");if(i){if(i.dims.length!==4)throw new Error('Input "attention_bias" must have 4 dimensions');if(i.dims[0]!==l||i.dims[1]!==r.numHeads||i.dims[2]!==c||i.dims[3]!==A)throw new Error('Expect "attention_bias" shape (batch_size, num_heads, sequence_length, total_sequence_length)')}return{batchSize:l,sequenceLength:c,pastSequenceLength:b,kvSequenceLength:f,totalSequenceLength:A,maxSequenceLength:g,inputHiddenSize:p,hiddenSize:d,vHiddenSize:_,headSize:Math.floor(d/r.numHeads),vHeadSize:Math.floor(_/r.numHeads),numHeads:r.numHeads,isUnidirectional:!1,pastPresentShareBuffer:!1,maskFilterValue:r.maskFilterValue,maskType:y,scale:r.scale,broadcastResPosBias:!1,passPastInKv:!1,qkvFormat:1}},ma=(e,r,t)=>r&&e?`
      let total_sequence_length_input = u32(${r.getByOffset("0")});
      let present_sequence_length = max(total_sequence_length_input, uniforms.past_sequence_length);
      let is_subsequent_prompt: bool = sequence_length > 1 && sequence_length != total_sequence_length_input;
      let is_first_prompt: bool = is_subsequent_prompt == false && sequence_length == total_sequence_length_input;
      total_sequence_length = u32(${e==null?void 0:e.getByOffset("batchIdx")}) + 1;
      var past_sequence_length: u32 = 0;
      if (is_first_prompt == false) {
        past_sequence_length = total_sequence_length - sequence_length;
      }
       `:`
    ${t?"let past_sequence_length = uniforms.past_sequence_length":""};
    let present_sequence_length = total_sequence_length;
    `,dh=(e,r,t,s,o,n,a,i)=>{let l=or(a?1:n),c=64,p=n/l;p<c&&(c=32);let d=Math.ceil(n/l/c),u=[{type:12,data:r},{type:12,data:t},{type:12,data:s},{type:12,data:o},{type:12,data:p},{type:12,data:d}],_=Sr(e.dataType,l),f=jr(1,l),b=["type"];a&&b.push("type"),i&&b.push("type");let A=g=>{let y=at("x",e.dataType,e.dims,l),C=[y],x=a?Oe("seq_lens",a.dataType,a.dims):void 0;x&&C.push(x);let M=i?Oe("total_sequence_length_input",i.dataType,i.dims):void 0;M&&C.push(M);let T=jr(e.dataType),v=[{name:"batch_size",type:"u32"},{name:"num_heads",type:"u32"},{name:"past_sequence_length",type:"u32"},{name:"sequence_length",type:"u32"},{name:"total_sequence_length",type:"u32"},{name:"elements_per_thread",type:"u32"}];return`
  var<workgroup> thread_max: array<f32, ${c}>;
  var<workgroup> thread_sum: array<f32, ${c}>;
  ${g.registerUniforms(v).declareVariables(...C)}
  ${g.mainStart([c,1,1])}
    let batchIdx = workgroup_id.z / uniforms.num_heads;
    let headIdx = workgroup_id.z % uniforms.num_heads;
    let sequence_length = uniforms.sequence_length;
    var total_sequence_length = uniforms.total_sequence_length;
    ${ma(x,M,!1)}
    let local_offset = local_idx * uniforms.elements_per_thread;
    let offset = (global_idx / ${c}) * uniforms.total_sequence_length + local_offset;
    let seq_causal_length = ${a?"u32(past_sequence_length + workgroup_id.y + 1)":"total_sequence_length"};
    var thread_max_vector = ${f}(-3.402823e+38f);
    for (var i: u32 = 0; i < uniforms.elements_per_thread && i + local_offset < seq_causal_length; i++) {
      thread_max_vector = max(${f}(x[offset + i]), thread_max_vector);
    }
    thread_max[local_idx] = ${(()=>{switch(l){case 1:return"thread_max_vector";case 2:return"max(thread_max_vector.x, thread_max_vector.y)";case 4:return"max(max(thread_max_vector.x, thread_max_vector.y), max(thread_max_vector.z, thread_max_vector.w))";default:throw new Error(`Unsupported components: ${l}`)}})()};
    workgroupBarrier();

    var max_value =  f32(-3.402823e+38f);
    for (var i = 0u; i < ${c}; i++) {
      max_value = max(thread_max[i], max_value);
    }

    var sum_vector = ${f}(0);
    for (var i: u32 = 0; i < uniforms.elements_per_thread && i + local_offset < seq_causal_length; i++) {
      sum_vector += exp(${f}(x[offset + i]) - max_value);
    }
    thread_sum[local_idx] = ${(()=>{switch(l){case 1:return"sum_vector";case 2:return"sum_vector.x + sum_vector.y";case 4:return"sum_vector.x + sum_vector.y + sum_vector.z + sum_vector.w";default:throw new Error(`Unsupported components: ${l}`)}})()};
    workgroupBarrier();

    var sum: f32 = 0;
    for (var i = 0u; i < ${c}; i++) {
      sum += thread_sum[i];
    }

    if (sum == 0) {
      for (var i: u32 = 0; i < uniforms.elements_per_thread && i + local_offset < seq_causal_length; i++) {
        x[offset + i] = ${y.type.value}(${T}(1.0) / ${T}(seq_causal_length));
      }
    } else {
      for (var i: u32 = 0; i < uniforms.elements_per_thread && i + local_offset < seq_causal_length; i++) {
        var f32input = ${f}(x[offset + i]);
        x[offset + i] = ${y.type.value}(exp(f32input - max_value) / sum);
      }
    }
      ${a?`
        for (var total_seq_id: u32 = seq_causal_length; total_seq_id + local_offset < uniforms.total_sequence_length; total_seq_id++) {
          x[offset + total_seq_id] = ${y.type.value}(${T}(0));
        }`:""};
  }`};return{name:"AttentionProbsSoftmax",shaderCache:{hint:`${c};${_};${l}`,inputDependencies:b},getShaderSource:A,getRunData:()=>({outputs:[],dispatchGroup:{x:1,y:o,z:r*t},programUniforms:u})}},ph=(e,r,t,s,o,n,a,i,l)=>{let c=a+n.kvSequenceLength,p=[n.batchSize,n.numHeads,n.sequenceLength,c],d=e>1&&s,u=n.kvNumHeads?n.kvNumHeads:n.numHeads,_=d?[n.batchSize,u,c,n.headSize]:void 0,f=n.nReps?n.nReps:1,b=n.scale===0?1/Math.sqrt(n.headSize):n.scale,A=or(n.headSize),g=n.headSize/A,y=12,C={x:Math.ceil(c/y),y:Math.ceil(n.sequenceLength/y),z:n.batchSize*n.numHeads},x=[{type:12,data:n.sequenceLength},{type:12,data:g},{type:12,data:c},{type:12,data:n.numHeads},{type:12,data:n.headSize},{type:1,data:b},{type:12,data:a},{type:12,data:n.kvSequenceLength},{type:12,data:f}],M=d&&s&&Pe.size(s.dims)>0,T=["type","type"];M&&T.push("type"),o&&T.push("type"),i&&T.push("type"),l&&T.push("type");let v=[{dims:p,dataType:r.dataType,gpuDataType:0}];d&&v.push({dims:_,dataType:r.dataType,gpuDataType:0});let P=F=>{let D=Oe("q",r.dataType,r.dims,A),K=Oe("key",t.dataType,t.dims,A),U=[D,K];if(M){let he=Oe("past_key",s.dataType,s.dims,A);U.push(he)}o&&U.push(Oe("attention_bias",o.dataType,o.dims));let j=i?Oe("seq_lens",i.dataType,i.dims):void 0;j&&U.push(j);let ne=l?Oe("total_sequence_length_input",l.dataType,l.dims):void 0;ne&&U.push(ne);let q=at("output",r.dataType,p),te=[q];d&&te.push(at("present_key",r.dataType,_,A));let Z=jr(1,A),ae=[{name:"M",type:"u32"},{name:"K",type:"u32"},{name:"N",type:"u32"},{name:"num_heads",type:"u32"},{name:"head_size",type:"u32"},{name:"alpha",type:"f32"},{name:"past_sequence_length",type:"u32"},{name:"kv_sequence_length",type:"u32"},{name:"n_reps",type:"u32"}];return`
  const TILE_SIZE = ${y}u;

  var<workgroup> tileQ: array<${D.type.storage}, ${y*y}>;
  var<workgroup> tileK: array<${D.type.storage}, ${y*y}>;
  ${F.registerUniforms(ae).declareVariables(...U,...te)}
  ${F.mainStart([y,y,1])}
    // x holds the N and y holds the M
    let headIdx = workgroup_id.z % uniforms.num_heads;
    let kvHeadIdx = ${f===1?"headIdx":"headIdx / uniforms.n_reps"};
    let kv_num_heads = ${f===1?"uniforms.num_heads":"uniforms.num_heads / uniforms.n_reps"};
    let batchIdx = workgroup_id.z / uniforms.num_heads;
    let m = workgroup_id.y * TILE_SIZE;
    let n = workgroup_id.x * TILE_SIZE;
    let sequence_length = uniforms.M;
    var total_sequence_length = uniforms.N;
    ${ma(j,ne,!0)}
    let absKvHeadIdx = batchIdx * kv_num_heads + kvHeadIdx;
    let qOffset = workgroup_id.z * uniforms.M * uniforms.K + m * uniforms.K;
    ${M&&d?"let pastKeyOffset = absKvHeadIdx * uniforms.past_sequence_length * uniforms.K;":""};
    let kOffset = absKvHeadIdx * uniforms.kv_sequence_length * uniforms.K;
    ${d?"let presentKeyOffset = absKvHeadIdx * uniforms.N * uniforms.K;":""}
    var value = ${Z}(0);
    for (var w: u32 = 0u; w < uniforms.K; w += TILE_SIZE) {
      if (global_id.y < uniforms.M && w + local_id.x < uniforms.K) {
        tileQ[TILE_SIZE * local_id.y + local_id.x] = q[qOffset + local_id.y * uniforms.K + w + local_id.x];
      }
      if (n + local_id.y < uniforms.N && w + local_id.x < uniforms.K) {
        var idx = TILE_SIZE * local_id.y + local_id.x;
      ${M&&d?`
              if (n + local_id.y < past_sequence_length) {
                tileK[idx] = past_key[pastKeyOffset + (n + local_id.y) * uniforms.K + w + local_id.x];
              } else if (n + local_id.y - past_sequence_length < uniforms.kv_sequence_length) {
                tileK[idx] = key[kOffset + (n + local_id.y - past_sequence_length) * uniforms.K + w + local_id.x];
              }`:`
          if (n + local_id.y < uniforms.kv_sequence_length) {
            tileK[idx] = key[kOffset + (n + local_id.y) * uniforms.K + w + local_id.x];
          }`}
      ${d?`if (n + local_id.y < present_sequence_length) {
        present_key[presentKeyOffset + (n + local_id.y) * uniforms.K + w + local_id.x] = tileK[idx];
      }`:""}
      }
      workgroupBarrier();

      for (var k: u32 = 0u; k < TILE_SIZE && w+k < uniforms.K; k++) {
          value += ${Z}(tileQ[TILE_SIZE * local_id.y + k] * tileK[TILE_SIZE * local_id.x + k]);
      }

      workgroupBarrier();
    }

    if (global_id.y < uniforms.M && global_id.x < total_sequence_length) {
      let headOffset = workgroup_id.z * uniforms.M * uniforms.N;
      let outputIdx = headOffset + global_id.y * uniforms.N + global_id.x;
      var sum: f32 = ${(()=>{switch(A){case 1:return"value";case 2:return"value.x + value.y";case 4:return"value.x + value.y + value.z + value.w";default:throw new Error(`Unsupported components: ${A}`)}})()};
        output[outputIdx] = ${q.type.value} (sum * uniforms.alpha) + ${o?"attention_bias[outputIdx]":"0.0"};
    }
  }`};return{name:"AttentionProbs",shaderCache:{hint:`${A};${o!==void 0};${s!==void 0};${e}`,inputDependencies:T},getRunData:()=>({outputs:v,dispatchGroup:C,programUniforms:x}),getShaderSource:P}},hh=(e,r,t,s,o,n,a=void 0,i=void 0)=>{let l=n+o.kvSequenceLength,c=o.nReps?o.nReps:1,p=o.vHiddenSize*c,d=e>1&&s,u=o.kvNumHeads?o.kvNumHeads:o.numHeads,_=d?[o.batchSize,u,l,o.headSize]:void 0,f=[o.batchSize,o.sequenceLength,p],b=12,A={x:Math.ceil(o.vHeadSize/b),y:Math.ceil(o.sequenceLength/b),z:o.batchSize*o.numHeads},g=[{type:12,data:o.sequenceLength},{type:12,data:l},{type:12,data:o.vHeadSize},{type:12,data:o.numHeads},{type:12,data:o.headSize},{type:12,data:p},{type:12,data:n},{type:12,data:o.kvSequenceLength},{type:12,data:c}],y=d&&s&&Pe.size(s.dims)>0,C=["type","type"];y&&C.push("type"),a&&C.push("type"),i&&C.push("type");let x=[{dims:f,dataType:r.dataType,gpuDataType:0}];d&&x.push({dims:_,dataType:r.dataType,gpuDataType:0});let M=T=>{let v=Oe("probs",r.dataType,r.dims),P=Oe("v",t.dataType,t.dims),F=[v,P];y&&F.push(Oe("past_value",s.dataType,s.dims));let D=a?Oe("seq_lens",a.dataType,a.dims):void 0;a&&F.push(D);let K=i?Oe("total_sequence_length_input",i.dataType,i.dims):void 0;i&&F.push(K);let U=[at("output",r.dataType,f)];d&&U.push(at("present_value",r.dataType,_));let j=[{name:"M",type:"u32"},{name:"K",type:"u32"},{name:"N",type:"u32"},{name:"num_heads",type:"u32"},{name:"head_size",type:"u32"},{name:"v_hidden_size",type:"u32"},{name:"past_sequence_length",type:"u32"},{name:"kv_sequence_length",type:"u32"},{name:"n_reps",type:"u32"}];return`
  const TILE_SIZE = ${b}u;
  var<workgroup> tileQ: array<${v.type.value}, ${b*b}>;
  var<workgroup> tileV: array<${v.type.value}, ${b*b}>;
  ${T.registerUniforms(j).declareVariables(...F,...U)}
  ${T.mainStart([b,b,1])}
   let headIdx = workgroup_id.z % uniforms.num_heads;
   let batchIdx = workgroup_id.z / uniforms.num_heads;
   let kvHeadIdx = ${c===1?"headIdx":"headIdx / uniforms.n_reps"};
   let kv_num_heads = ${c===1?"uniforms.num_heads":"uniforms.num_heads / uniforms.n_reps"};
   let m = global_id.y;
   let n = global_id.x;
   let sequence_length = uniforms.M;
   var total_sequence_length = uniforms.K;
   ${ma(D,K,!0)}
   let offsetA = workgroup_id.z * uniforms.M * uniforms.K + m * uniforms.K;
   let absKvHeadIdx = batchIdx * kv_num_heads + kvHeadIdx; // kvHeadIdx is relative to the batch
   ${y&&d?"let pastValueOffset = absKvHeadIdx * uniforms.N * uniforms.past_sequence_length + n;":""};
   let vOffset = absKvHeadIdx * uniforms.N * uniforms.kv_sequence_length + n;
   ${d?"let presentValueOffset = absKvHeadIdx * uniforms.N * uniforms.K + n;":""}
   var value = ${v.type.storage}(0);
   for (var w: u32 = 0u; w < uniforms.K; w += TILE_SIZE) {
      if (m < uniforms.M && w + local_id.x < uniforms.K) {
        tileQ[TILE_SIZE * local_id.y + local_id.x] = probs[offsetA + w + local_id.x];
      }
      if (n < uniforms.N && w + local_id.y < uniforms.K) {
        var idx = TILE_SIZE * local_id.y + local_id.x;
        ${y&&d?`
        if (w + local_id.y < past_sequence_length) {
          tileV[idx] = past_value[pastValueOffset + (w + local_id.y) * uniforms.N];
        } else if (w + local_id.y - past_sequence_length < uniforms.kv_sequence_length) {
          tileV[idx] = v[vOffset + (w + local_id.y - past_sequence_length) * uniforms.N];
        }
      `:`
            if (w + local_id.y < uniforms.kv_sequence_length) {
              tileV[idx] = v[vOffset + (w + local_id.y) * uniforms.N];
            }`}
        ${d?`
            if (w + local_id.y < present_sequence_length) {
          present_value[presentValueOffset + (w + local_id.y) * uniforms.N] = tileV[idx];
        }`:""}
      }
     workgroupBarrier();
     for (var k: u32 = 0u; k < TILE_SIZE && w+k < total_sequence_length; k++) {
       value += tileQ[TILE_SIZE * local_id.y + k] * tileV[TILE_SIZE * k + local_id.x];
     }
     workgroupBarrier();
   }

   // we need to transpose output from BNSH_v to BSND_v
   if (m < uniforms.M && n < uniforms.N) {
     let outputIdx = batchIdx * uniforms.M * uniforms.v_hidden_size + m * uniforms.v_hidden_size
       + headIdx * uniforms.N + n;
     output[outputIdx] = value;
   }
  }`};return{name:"AttentionScore",shaderCache:{hint:`${s!==void 0};${e}`,inputDependencies:C},getRunData:()=>({outputs:x,dispatchGroup:A,programUniforms:g}),getShaderSource:M}},wo=(e,r,t,s,o,n,a,i,l,c,p=void 0,d=void 0)=>{let u=Math.min(e.outputCount,1+(a?1:0)+(i?1:0)),_=u>1?c.pastSequenceLength:0,f=_+c.kvSequenceLength,b=l&&Pe.size(l.dims)>0?l:void 0,A=[r,t];u>1&&a&&Pe.size(a.dims)>0&&A.push(a),b&&A.push(b),p&&A.push(p),d&&A.push(d);let g=e.compute(ph(u,r,t,a,b,c,_,p,d),{inputs:A,outputs:u>1?[-1,1]:[-1]})[0];e.compute(dh(g,c.batchSize,c.numHeads,_,c.sequenceLength,f,p,d),{inputs:p&&d?[g,p,d]:[g],outputs:[]});let y=[g,s];u>1&&i&&Pe.size(i.dims)>0&&y.push(i),p&&y.push(p),d&&y.push(d),e.compute(hh(u,g,s,i,c,_,p,d),{inputs:y,outputs:u>1?[0,2]:[0]})},mh=(e,r)=>{let t=[r.batchSize,r.numHeads,r.sequenceLength,r.headSize],s=r.sequenceLength,o=r.inputHiddenSize,n=r.headSize,a=12,i={x:Math.ceil(r.headSize/a),y:Math.ceil(r.sequenceLength/a),z:r.batchSize*r.numHeads},l=[e.inputs[0],e.inputs[1],e.inputs[2]],c=[{type:12,data:s},{type:12,data:o},{type:12,data:n},{type:12,data:r.numHeads},{type:12,data:r.headSize},{type:12,data:r.hiddenSize},{type:12,data:r.hiddenSize+r.hiddenSize+r.vHiddenSize}],p=d=>{let u=at("output_q",l[0].dataType,t),_=at("output_k",l[0].dataType,t),f=at("output_v",l[0].dataType,t),b=Oe("input",l[0].dataType,l[0].dims),A=Oe("weight",l[1].dataType,l[1].dims),g=Oe("bias",l[2].dataType,l[2].dims),y=b.type.storage,C=[{name:"M",type:"u32"},{name:"K",type:"u32"},{name:"N",type:"u32"},{name:"num_heads",type:"u32"},{name:"head_size",type:"u32"},{name:"hidden_size",type:"u32"},{name:"ldb",type:"u32"}];return`
  const TILE_SIZE = ${a}u;
  var<workgroup> tileInput: array<${y}, ${a*a}>;
  var<workgroup> tileWeightQ: array<${y}, ${a*a}>;
  var<workgroup> tileWeightK: array<${y}, ${a*a}>;
  var<workgroup> tileWeightV: array<${y}, ${a*a}>;
  ${d.registerUniforms(C).declareVariables(b,A,g,u,_,f)}
  ${d.mainStart([a,a,1])}
    let batchIndex = workgroup_id.z / uniforms.num_heads;
    let headNumber = workgroup_id.z % uniforms.num_heads;
    let m = global_id.y;
    let n = global_id.x;

    let inputOffset = batchIndex * (uniforms.M * uniforms.K) + m * uniforms.K;
    let biasOffsetQ = headNumber * uniforms.head_size;
    let biasOffsetK = uniforms.hidden_size + biasOffsetQ;
    let biasOffsetV = uniforms.hidden_size + biasOffsetK;

    var valueQ = ${y}(0);
    var valueK = ${y}(0);
    var valueV = ${y}(0);
    for (var w: u32 = 0u; w < uniforms.K; w += TILE_SIZE) {
      if (m < uniforms.M && w + local_id.x < uniforms.K) {
        tileInput[TILE_SIZE * local_id.y + local_id.x] = input[inputOffset + w + local_id.x];
      }
      if (n < uniforms.N && w + local_id.y < uniforms.K) {
        let offset = n + (w + local_id.y) * uniforms.ldb;
        tileWeightQ[TILE_SIZE * local_id.y + local_id.x] = weight[biasOffsetQ + offset];
        tileWeightK[TILE_SIZE * local_id.y + local_id.x] = weight[biasOffsetK + offset];
        tileWeightV[TILE_SIZE * local_id.y + local_id.x] = weight[biasOffsetV + offset];
      }
      workgroupBarrier();
      for (var k: u32 = 0u; k<TILE_SIZE && w+k < uniforms.K; k++) {
        let inputTileOffset = TILE_SIZE * local_id.y + k;
        let weightTileOffset = TILE_SIZE * k + local_id.x;
        valueQ += tileInput[inputTileOffset] * tileWeightQ[weightTileOffset];
        valueK += tileInput[inputTileOffset] * tileWeightK[weightTileOffset];
        valueV += tileInput[inputTileOffset] * tileWeightV[weightTileOffset];
      }

      workgroupBarrier();
    }

    let headOffset = (m * uniforms.N + n) % uniforms.head_size;
    valueQ += bias[headOffset + biasOffsetQ];
    valueK += bias[headOffset + biasOffsetK];
    valueV += bias[headOffset + biasOffsetV];

    let offset = workgroup_id.z * uniforms.M * uniforms.N;
    if (m < uniforms.M && n < uniforms.N) {
      let outputIdx = offset + m * uniforms.N + n;
      output_q[outputIdx] = valueQ;
      output_k[outputIdx] = valueK;
      output_v[outputIdx] = valueV;
    }
  }`};return e.compute({name:"AttentionPrepare",shaderCache:{inputDependencies:["type","type","type"]},getRunData:()=>({outputs:[{dims:t,dataType:e.inputs[0].dataType,gpuDataType:0},{dims:t,dataType:e.inputs[0].dataType,gpuDataType:0},{dims:t,dataType:e.inputs[0].dataType,gpuDataType:0}],dispatchGroup:i,programUniforms:c}),getShaderSource:p},{inputs:l,outputs:[-1,-1,-1]})},fh=(e,r)=>{let t=ch(e.inputs,r),[s,o,n]=mh(e,t);return wo(e,s,o,n,e.inputs[4],void 0,void 0,void 0,e.inputs[5],t)}}),_h,gh,wh,Mh,Xx=Ue(()=>{bs(),Mt(),Pt(),ur(),Ct(),_h=(e,r)=>{if(!e||e.length!==5)throw new Error("BatchNormalization requires 5 inputs");let t=(s,o,n)=>{let a=o.length;if(a!==s.length)throw new Error(`${n}: num dimensions != ${a}`);o.forEach((i,l)=>{if(i!==s[l])throw new Error(`${n}: dim[${l}] do not match`)})};if(e[0].dims.length>1){let s=r.format==="NHWC"?r.spatial?e[0].dims.slice(-1):e[0].dims.slice(-1).concat(e[0].dims.slice(1,e[0].dims.length-1)):e[0].dims.slice(1,r.spatial?2:void 0);t(e[1].dims,s,"Invalid input scale"),t(e[2].dims,s,"Invalid input B"),t(e[3].dims,s,"Invalid input mean"),t(e[4].dims,s,"Invalid input var")}else t(e[1].dims,[1],"Invalid input scale"),t(e[2].dims,[1],"Invalid input B"),t(e[3].dims,[1],"Invalid input mean"),t(e[4].dims,[1],"Invalid input var")},gh=(e,r)=>{let{epsilon:t,spatial:s,format:o}=r,n=e[0].dims,a=s?or(n[n.length-1]):1,i=o==="NHWC"&&n.length>1?a:1,l=Pe.size(n)/a,c=s,p=c?n.length:n,d=Oe("x",e[0].dataType,e[0].dims,a),u=Oe("scale",e[1].dataType,e[1].dims,i),_=Oe("bias",e[2].dataType,e[2].dims,i),f=Oe("inputMean",e[3].dataType,e[3].dims,i),b=Oe("inputVar",e[4].dataType,e[4].dims,i),A=at("y",e[0].dataType,p,a),g=()=>{let C="";if(s)C=`let cOffset = ${n.length===1?"0u":o==="NHWC"?`outputIndices[${n.length-1}] / ${a}`:"outputIndices[1]"};`;else if(o==="NCHW")C=`
            ${A.indicesSet("outputIndices","0","0")}
            let cOffset = ${A.indicesToOffset("outputIndices")};`;else{C=`var cIndices = ${u.type.indices}(0);
                       cIndices[0] = outputIndices[${n.length-1}];`;for(let x=1;x<u.rank;x++)C+=`cIndices[${x}] = outputIndices[${x}];`;C+=`let cOffset = ${u.indicesToOffset("cIndices")};`}return C},y=C=>`
  const epsilon = ${t};
  ${C.registerUniform("outputSize","u32").declareVariables(d,u,_,f,b,A)}
  ${C.mainStart()}
  ${C.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.outputSize")}
    var outputIndices = ${A.offsetToIndices(`global_idx * ${a}`)};
    ${g()}
    let scale = ${u.getByOffset("cOffset")};
    let bias = ${_.getByOffset("cOffset")};
    let inputMean = ${f.getByOffset("cOffset")};
    let inputVar = ${b.getByOffset("cOffset")};
    let x = ${d.getByOffset("global_idx")};
    let value = (x - inputMean) * inverseSqrt(inputVar + epsilon) * scale + bias;
    ${A.setByOffset("global_idx","value")}
  }`;return{name:"BatchNormalization",shaderCache:{hint:`${r.epsilon}_${r.format}_${s}_${a}`,inputDependencies:c?["rank","type","type","type","type"]:void 0},getShaderSource:y,getRunData:()=>({outputs:[{dims:e[0].dims,dataType:e[0].dataType}],dispatchGroup:{x:Math.ceil(l/64)},programUniforms:c?[{type:12,data:l},...lt(n)]:[{type:12,data:l}]})}},wh=e=>jt(e),Mh=(e,r)=>{let{inputs:t,outputCount:s}=e,o=wh({...r,outputCount:s});if(Jt.webgpu.validateInputContent&&_h(t,o),r.trainingMode)throw new Error("BatchNormalization trainingMode is not supported yet.");e.compute(gh(t,o))}}),bh,yh,vh,Jx=Ue(()=>{Pt(),Ct(),bh=e=>{if(e[0].dims.length!==3)throw new Error("input should have 3 dimensions");if(![320,640,1280].includes(e[0].dims[2]))throw new Error("number of channels should be 320, 640 or 1280");if(e[1].dims.length!==1)throw new Error("bias is expected to have 1 dimensions");if(e[0].dims[2]!==e[1].dims[0])throw new Error("last dimension of input and bias are not the same")},yh=e=>{let r=e[0].dims,t=e[0].dims[2],s=Pe.size(r)/4,o=e[0].dataType,n=Oe("input",o,r,4),a=Oe("bias",o,[t],4),i=Oe("residual",o,r,4),l=at("output",o,r,4);return{name:"BiasAdd",getRunData:()=>({outputs:[{dims:r,dataType:e[0].dataType}],dispatchGroup:{x:Math.ceil(s/64)}}),getShaderSource:c=>`
  const channels = ${t}u / 4;
  ${c.declareVariables(n,a,i,l)}

  ${c.mainStart()}
    ${c.guardAgainstOutOfBoundsWorkgroupSizes(s)}
    let value = ${n.getByOffset("global_idx")}
      + ${a.getByOffset("global_idx % channels")} + ${i.getByOffset("global_idx")};
    ${l.setByOffset("global_idx","value")}
  }`}},vh=e=>{bh(e.inputs),e.compute(yh(e.inputs))}}),xh,Bt,Th,Eh,Ph,Ch,Sh,$h,Ah,kh,Ih,Fh,Oh,Dh,Lh,zh,Mo,Rh,fa,Bh,Nh,jh,Vh,Uh,Wh,Gh,Hh,Kh,qh,Qh,Xh,Jh,Yh,Zh,em,yl,tm,vl,xl,rm,sm,nm,om,am,im,Tl=Ue(()=>{Mt(),Pt(),ur(),Ct(),xh=(e,r,t,s,o,n,a)=>{let i=Math.ceil(r/4),l="";typeof o=="string"?l=`${o}(a)`:l=o("a");let c=Oe("inputData",t,[i],4),p=at("outputData",s,[i],4),d=[{name:"vec_size",type:"u32"}];return a&&d.push(...a),`
      ${e.registerUniforms(d).declareVariables(c,p)}

  ${n??""}

  ${e.mainStart()}
    ${e.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.vec_size")}

    let a = ${c.getByOffset("global_idx")};
    ${p.setByOffset("global_idx",l)}
  }`},Bt=(e,r,t,s,o,n=e.dataType,a,i)=>{let l=[{type:12,data:Math.ceil(Pe.size(e.dims)/4)}];return a&&l.push(...a),{name:r,shaderCache:{hint:o,inputDependencies:["type"]},getShaderSource:c=>xh(c,Pe.size(e.dims),e.dataType,n,t,s,i),getRunData:c=>({outputs:[{dims:e.dims,dataType:n}],dispatchGroup:{x:Math.ceil(Pe.size(c[0].dims)/64/4)},programUniforms:l})}},Th=e=>{e.compute(Bt(e.inputs[0],"Abs","abs"))},Eh=e=>{e.compute(Bt(e.inputs[0],"Acos","acos"))},Ph=e=>{e.compute(Bt(e.inputs[0],"Acosh","acosh"))},Ch=e=>{e.compute(Bt(e.inputs[0],"Asin","asin"))},Sh=e=>{e.compute(Bt(e.inputs[0],"Asinh","asinh"))},$h=e=>{e.compute(Bt(e.inputs[0],"Atan","atan"))},Ah=e=>{e.compute(Bt(e.inputs[0],"Atanh","atanh"))},kh=e=>jt(e),Ih=(e,r)=>{let t;switch(r.to){case 10:t="vec4<f16>";break;case 1:t="vec4<f32>";break;case 12:t="vec4<u32>";break;case 6:t="vec4<i32>";break;case 9:t="vec4<bool>";break;default:throw new RangeError(`not supported type (specified in attribute 'to' from 'Cast' operator): ${r.to}`)}e.compute(Bt(e.inputs[0],"Cast",t,void 0,r.cacheKey,r.to))},Fh=e=>{let r,t,s=e.length>=2&&e[1].data!==0,o=e.length>=3&&e[2].data!==0;switch(e[0].dataType){case 1:r=s?e[1].getFloat32Array()[0]:-34028234663852886e22,t=o?e[2].getFloat32Array()[0]:34028234663852886e22;break;case 10:r=s?e[1].getUint16Array()[0]:64511,t=o?e[2].getUint16Array()[0]:31743;break;default:throw new Error("Unsupport data type")}return jt({min:r,max:t})},Oh=(e,r)=>{let t=r||Fh(e.inputs),s=jr(e.inputs[0].dataType);e.compute(Bt(e.inputs[0],"Clip",o=>`clamp(${o}, vec4<${s}>(uniforms.min), vec4<${s}>(uniforms.max))`,void 0,t.cacheKey,void 0,[{type:e.inputs[0].dataType,data:t.min},{type:e.inputs[0].dataType,data:t.max}],[{name:"min",type:s},{name:"max",type:s}]),{inputs:[0]})},Dh=e=>{e.compute(Bt(e.inputs[0],"Ceil","ceil"))},Lh=e=>{e.compute(Bt(e.inputs[0],"Cos","cos"))},zh=e=>{e.compute(Bt(e.inputs[0],"Cosh","cosh"))},Mo=e=>jt(e),Rh=(e,r)=>{let t=jr(e.inputs[0].dataType);e.compute(Bt(e.inputs[0],"Elu",s=>`elu_vf32(${s})`,`
  const elu_alpha_ = ${t}(${r.alpha});

  fn elu_f32(a: ${t}) -> ${t} {
  return select((exp(a) - 1.0) * elu_alpha_, a, a >= 0.0);
  }

  fn elu_vf32(v: vec4<${t}>) -> vec4<${t}> {
  return vec4(elu_f32(v.x), elu_f32(v.y), elu_f32(v.z), elu_f32(v.w));
  }`,r.cacheKey))},fa=(e="f32")=>`
const r0: ${e} = 0.3275911;
const r1: ${e} = 0.254829592;
const r2: ${e} = -0.284496736;
const r3: ${e} = 1.421413741;
const r4: ${e} = -1.453152027;
const r5: ${e} = 1.061405429;

fn erf_vf32(v: vec4<${e}>) -> vec4<${e}> {
  let absv = abs(v);
  let x = 1.0 / (1.0 + r0 * absv);
  return sign(v) * (1.0 - ((((r5 * x + r4) * x + r3) * x + r2) * x + r1) * x * exp(-absv * absv));
}`,Bh=e=>{let r=jr(e.inputs[0].dataType);e.compute(Bt(e.inputs[0],"Erf",t=>`erf_vf32(${t})`,fa(r)))},Nh=e=>{e.compute(Bt(e.inputs[0],"Exp","exp"))},jh=e=>{e.compute(Bt(e.inputs[0],"Floor","floor"))},Vh=e=>{let r=jr(e.inputs[0].dataType);e.compute(Bt(e.inputs[0],"Gelu",t=>`0.5 * ${t} * (1.0 + erf_vf32(${t} * 0.7071067811865475))`,fa(r)))},Uh=(e,r)=>{let t=jr(e.inputs[0].dataType);e.compute(Bt(e.inputs[0],"LeakyRelu",s=>`select(leaky_relu_alpha_ * ${s}, ${s}, ${s} >= vec4<${t}>(0.0))`,`const leaky_relu_alpha_ = ${t}(${r.alpha});`,r.cacheKey))},Wh=e=>{e.compute(Bt(e.inputs[0],"Not",r=>`!${r}`))},Gh=e=>{e.compute(Bt(e.inputs[0],"Neg",r=>`-${r}`))},Hh=e=>{e.compute(Bt(e.inputs[0],"Reciprocal",r=>`1.0/${r}`))},Kh=e=>{let r=jr(e.inputs[0].dataType);e.compute(Bt(e.inputs[0],"Relu",t=>`select(vec4<${r}>(0.0), ${t}, ${t} > vec4<${r}>(0.0))`))},qh=e=>{e.compute(Bt(e.inputs[0],"Sigmoid",r=>`(1.0 / (1.0 + exp(-${r})))`))},Qh=e=>jt(e),Xh=(e,r)=>{let t=jr(e.inputs[0].dataType);e.compute(Bt(e.inputs[0],"HardSigmoid",s=>`max(vec4<${t}>(0.0), min(vec4<${t}>(1.0), ${r.alpha} * ${s} + vec4<${t}>(${r.beta})))`,void 0,r.cacheKey))},Jh=e=>{e.compute(Bt(e.inputs[0],"Sin","sin"))},Yh=e=>{e.compute(Bt(e.inputs[0],"Sinh","sinh"))},Zh=e=>{e.compute(Bt(e.inputs[0],"Sqrt","sqrt"))},em=e=>{e.compute(Bt(e.inputs[0],"Tan","tan"))},yl=e=>`sign(${e}) * (1 - exp(-2 * abs(${e}))) / (1 + exp(-2 * abs(${e})))`,tm=e=>{e.compute(Bt(e.inputs[0],"Tanh",yl))},vl=(e="f32")=>`
const fast_gelu_a: ${e} = 0.5;
const fast_gelu_b: ${e} = 0.7978845608028654;
const fast_gelu_c: ${e} = 0.035677408136300125;

fn tanh_v(v: vec4<${e}>) -> vec4<${e}> {
  return ${yl("v")};
}
`,xl=e=>`(fast_gelu_a + fast_gelu_a * tanh_v(${e} * (fast_gelu_c * ${e} * ${e} + fast_gelu_b))) * ${e}`,rm=e=>{let r=jr(e.inputs[0].dataType);e.compute(Bt(e.inputs[0],"FastGelu",xl,vl(r),void 0,e.inputs[0].dataType))},sm=(e,r)=>{let t=jr(e.inputs[0].dataType);return e.compute(Bt(e.inputs[0],"ThresholdedRelu",s=>`select(vec4<${t}>(0.0), ${s}, ${s} > thresholded_relu_alpha_)`,`const thresholded_relu_alpha_ = vec4<${t}>(${r.alpha});`,r.cacheKey)),0},nm=e=>{e.compute(Bt(e.inputs[0],"Log","log"))},om=(e,r)=>`
const alpha = vec4<${e}>(${r});
const one = ${e}(1.0);
const zero = ${e}(0.0);

fn quick_gelu_impl(x: vec4<${e}>) -> vec4<${e}> {
  let v = x *alpha;
  var x1 : vec4<${e}>;
  for (var i = 0; i < 4; i = i + 1) {
    if (v[i] >= zero) {
      x1[i] = one / (one + exp(-v[i]));
    } else {
      x1[i] = one - one / (one + exp(v[i]));
    }
  }
  return x * x1;
}
`,am=e=>`quick_gelu_impl(${e})`,im=(e,r)=>{let t=jr(e.inputs[0].dataType);e.compute(Bt(e.inputs[0],"QuickGelu",am,om(t,r.alpha),r.cacheKey,e.inputs[0].dataType))}}),lm,um,cm,Yx=Ue(()=>{Pt(),Ct(),Tl(),lm=e=>{if(e[0].dims.length!==3)throw new Error("input should have 3 dimensions");if(![2560,5120,10240].includes(e[0].dims[2]))throw new Error("hidden state should be 2560, 5120 or 10240");if(e[1].dims.length!==1)throw new Error("bias is expected to have 1 dimensions");if(e[0].dims[2]!==e[1].dims[0])throw new Error("last dimension of input and bias are not the same")},um=e=>{let r=e[0].dims.slice();r[2]=r[2]/2;let t=Oe("input",e[0].dataType,e[0].dims,4),s=Oe("bias",e[0].dataType,[e[0].dims[2]],4),o=at("output",e[0].dataType,r,4),n=Pe.size(r)/4,a=Sr(e[0].dataType);return{name:"BiasSplitGelu",getRunData:()=>({outputs:[{dims:r,dataType:e[0].dataType}],dispatchGroup:{x:Math.ceil(n/64)}}),getShaderSource:i=>`
  const M_SQRT2 = sqrt(2.0);
  const halfChannels = ${e[0].dims[2]/4/2}u;

  ${i.declareVariables(t,s,o)}

  ${fa(a)}

  ${i.mainStart()}
    ${i.guardAgainstOutOfBoundsWorkgroupSizes(n)}
    let biasIdx = global_idx % halfChannels;
    let batchIndex = global_idx / halfChannels;
    let inputOffset = biasIdx + batchIndex * halfChannels * 2;
    let valueLeft = input[inputOffset] + bias[biasIdx];
    let valueRight = input[inputOffset + halfChannels] + bias[biasIdx + halfChannels];
    let geluRight = valueRight * 0.5 * (erf_vf32(valueRight / M_SQRT2) + 1);

    ${o.setByOffset("global_idx","valueLeft * geluRight")}
  }`}},cm=e=>{lm(e.inputs),e.compute(um(e.inputs))}}),dm,pm,Ps,hm,mm,fm,_m,gm,wm,Mm,bm,ym,vm,Zx=Ue(()=>{Mt(),Pt(),Ct(),dm=(e,r,t,s,o,n,a,i,l,c,p,d)=>{let u,_;typeof i=="string"?u=_=(y,C)=>`${i}((${y}),(${C}))`:typeof i=="function"?u=_=i:(u=i.scalar,_=i.vector);let f=at("outputData",p,s.length,4),b=Oe("aData",l,r.length,4),A=Oe("bData",c,t.length,4),g;if(o)if(n){let y=Pe.size(r)===1,C=Pe.size(t)===1,x=r.length>0&&r[r.length-1]%4===0,M=t.length>0&&t[t.length-1]%4===0;y||C?g=f.setByOffset("global_idx",_(y?`${b.type.value}(${b.getByOffset("0")}.x)`:b.getByOffset("global_idx"),C?`${A.type.value}(${A.getByOffset("0")}.x)`:A.getByOffset("global_idx"))):g=`
            let outputIndices = ${f.offsetToIndices("global_idx * 4u")};
            let offsetA = ${b.broadcastedIndicesToOffset("outputIndices",f)};
            let offsetB = ${A.broadcastedIndicesToOffset("outputIndices",f)};
            ${f.setByOffset("global_idx",_(a||x?b.getByOffset("offsetA / 4u"):`${b.type.value}(${b.getByOffset("offsetA / 4u")}[offsetA % 4u])`,a||M?A.getByOffset("offsetB / 4u"):`${A.type.value}(${A.getByOffset("offsetB / 4u")}[offsetB % 4u])`))}
          `}else g=f.setByOffset("global_idx",_(b.getByOffset("global_idx"),A.getByOffset("global_idx")));else{if(!n)throw new Error("no necessary to use scalar implementation for element-wise binary op implementation.");let y=(C,x,M="")=>{let T=`aData[indexA${x}][componentA${x}]`,v=`bData[indexB${x}][componentB${x}]`;return`
            let outputIndices${x} = ${f.offsetToIndices(`global_idx * 4u + ${x}u`)};
            let offsetA${x} = ${b.broadcastedIndicesToOffset(`outputIndices${x}`,f)};
            let offsetB${x} = ${A.broadcastedIndicesToOffset(`outputIndices${x}`,f)};
            let indexA${x} = offsetA${x} / 4u;
            let indexB${x} = offsetB${x} / 4u;
            let componentA${x} = offsetA${x} % 4u;
            let componentB${x} = offsetB${x} % 4u;
            ${C}[${x}] = ${M}(${u(T,v)});
          `};p===9?g=`
            var data = vec4<u32>(0);
            ${y("data",0,"u32")}
            ${y("data",1,"u32")}
            ${y("data",2,"u32")}
            ${y("data",3,"u32")}
            outputData[global_idx] = dot(vec4<u32>(0x1, 0x100, 0x10000, 0x1000000), vec4<u32>(data));`:g=`
            ${y("outputData[global_idx]",0)}
            ${y("outputData[global_idx]",1)}
            ${y("outputData[global_idx]",2)}
            ${y("outputData[global_idx]",3)}
          `}return`
        ${e.registerUniform("vec_size","u32").declareVariables(b,A,f)}

        ${d??""}

        ${e.mainStart()}
        ${e.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.vec_size")}
        ${g}
      }`},pm=(e,r,t,s,o,n,a=t.dataType)=>{let i=t.dims.map(b=>Number(b)??1),l=s.dims.map(b=>Number(b)??1),c=!Pe.areEqual(i,l),p=i,d=Pe.size(i),u=!1,_=!1,f=[c];if(c){let b=Gn.calcShape(i,l,!1);if(!b)throw new Error("Can't perform binary op on the given tensors");p=b.slice(),d=Pe.size(p);let A=Pe.size(i)===1,g=Pe.size(l)===1,y=i.length>0&&i[i.length-1]%4===0,C=l.length>0&&l[l.length-1]%4===0;f.push(A),f.push(g),f.push(y),f.push(C);let x=1;for(let M=1;M<p.length;M++){let T=i[i.length-M],v=l[l.length-M];if(T===v)x*=T;else break}x%4===0?(_=!0,u=!0):(A||g||y||C)&&(u=!0)}else u=!0;return f.push(u),{name:e,shaderCache:{hint:r+f.map(b=>b.toString()).join("_"),inputDependencies:["rank","rank"]},getShaderSource:b=>dm(b,i,l,p,u,c,_,o,t.dataType,s.dataType,a,n),getRunData:()=>({outputs:[{dims:p,dataType:a}],dispatchGroup:{x:Math.ceil(d/64/4)},programUniforms:[{type:12,data:Math.ceil(Pe.size(p)/4)},...lt(i,l,p)]})}},Ps=(e,r,t,s,o,n)=>{e.compute(pm(r,o??"",e.inputs[0],e.inputs[1],t,s,n))},hm=e=>{Ps(e,"Add",(r,t)=>`${r}+${t}`)},mm=e=>{Ps(e,"Div",(r,t)=>`${r}/${t}`)},fm=e=>{Ps(e,"Equal",{scalar:(r,t)=>`u32(${r}==${t})`,vector:(r,t)=>`vec4<u32>(${r}==${t})`},void 0,void 0,9)},_m=e=>{Ps(e,"Mul",(r,t)=>`${r}*${t}`)},gm=e=>{let r=Oe("input",e.inputs[0].dataType,e.inputs[0].dims).type.value;Ps(e,"Pow",{scalar:(t,s)=>`pow_custom(${t},${s})`,vector:(t,s)=>`pow_vector_custom(${t},${s})`},`
    fn pow_custom(a : ${r}, b : ${r}) -> ${r} {
      if (b == ${r}(0.0)) {
        return ${r}(1.0);
      } else if (a < ${r}(0.0) && f32(b) != floor(f32(b))) {
        return ${r}(pow(f32(a), f32(b))); // NaN
      }
      return select(sign(a), ${r}(1.0), round(f32(abs(b) % ${r}(2.0))) != 1.0) * ${r}(${r==="i32"?"round":""}(pow(f32(abs(a)), f32(b))));
    }
    fn pow_vector_custom(a : vec4<${r}>, b : vec4<${r}>) -> vec4<${r}> {
      // TODO: implement vectorized pow
      return vec4<${r}>(pow_custom(a.x, b.x), pow_custom(a.y, b.y), pow_custom(a.z, b.z), pow_custom(a.w, b.w));
    }
      `)},wm=e=>{Ps(e,"Sub",(r,t)=>`${r}-${t}`)},Mm=e=>{Ps(e,"Greater",{scalar:(r,t)=>`u32(${r}>${t})`,vector:(r,t)=>`vec4<u32>(${r}>${t})`},void 0,void 0,9)},bm=e=>{Ps(e,"Less",{scalar:(r,t)=>`u32(${r}<${t})`,vector:(r,t)=>`vec4<u32>(${r}<${t})`},void 0,void 0,9)},ym=e=>{Ps(e,"GreaterOrEqual",{scalar:(r,t)=>`u32(${r}>=${t})`,vector:(r,t)=>`vec4<u32>(${r}>=${t})`},void 0,void 0,9)},vm=e=>{Ps(e,"LessOrEqual",{scalar:(r,t)=>`u32(${r}<=${t})`,vector:(r,t)=>`vec4<u32>(${r}<=${t})`},void 0,void 0,9)}}),xm,Tm,Em,Pm,Cm,Sm,eT=Ue(()=>{Mt(),Pt(),ur(),Ct(),xm=(e,r)=>{if(!e||e.length<1)throw new Error("too few inputs");let t=0,s=e[t],o=s.dataType,n=s.dims.length;e.forEach((a,i)=>{if(i!==t){if(a.dataType!==o)throw new Error("input tensors should be one type");if(a.dims.length!==n)throw new Error("input tensors should have the same shape");a.dims.forEach((l,c)=>{if(c!==r&&l!==s.dims[c])throw new Error("non concat dimensions must match")})}})},Tm=(e,r)=>`
  fn calculateInputIndex(index: u32) -> u32 {
    let sizeInConcatAxis = array<u32, ${e}u>(${r});
    for (var i: u32 = 0u; i < ${e}; i += 1u ) {
      if (index < sizeInConcatAxis[i]) {
        return i;
      }
    }
    return ${e}u;
  }`,Em=(e,r)=>{let t=e.length,s=[];for(let o=0;o<t;++o){let n=r.setByOffset("global_idx",e[o].getByIndices("indices"));t===1?s.push(n):o===0?s.push(`if (inputIndex == ${o}u) { ${n} }`):o===t-1?s.push(`else { ${n} }`):s.push(`else if (inputIndex == ${o}) { ${n} }`)}return s.join(`
`)},Pm=(e,r,t,s)=>{let o=Pe.size(t),n=new Array(e.length),a=new Array(e.length),i=0,l=[],c=[],p=[{type:12,data:o}];for(let b=0;b<e.length;++b)i+=e[b].dims[r],n[b]=i,c.push(e[b].dims.length),a[b]=Oe(`input${b}`,s,c[b]),l.push("rank"),p.push({type:12,data:n[b]});for(let b=0;b<e.length;++b)p.push(...lt(e[b].dims));p.push(...lt(t));let d=at("output",s,t.length),u=d.indicesGet("indices",r),_=Array.from(Array(n.length).keys()).map(b=>`uniforms.sizeInConcatAxis${b}`).join(","),f=b=>`

  ${(()=>{b.registerUniform("outputSize","u32");for(let A=0;A<e.length;A++)b.registerUniform(`sizeInConcatAxis${A}`,"u32");return b.declareVariables(...a,d)})()}

  ${Tm(n.length,_)}

  ${b.mainStart()}
    ${b.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.outputSize")}

    var indices = ${d.offsetToIndices("global_idx")};

    let inputIndex = calculateInputIndex(${u});
    if (inputIndex != 0u) {
      let sizeInConcatAxis = array<u32, ${n.length}u>(${_});
      ${u} -= sizeInConcatAxis[inputIndex - 1u];
    }

    ${Em(a,d)}
  }`;return{name:"Concat",shaderCache:{hint:`${r}`,inputDependencies:l},getRunData:()=>({outputs:[{dims:t,dataType:s}],dispatchGroup:{x:Math.ceil(o/64)},programUniforms:p}),getShaderSource:f}},Cm=(e,r)=>{let t=e.inputs,s=t[0].dims,o=Pe.normalizeAxis(r.axis,s.length);xm(t,o);let n=s.slice();n[o]=t.reduce((i,l)=>i+(l.dims.length>o?l.dims[o]:0),0);let a=t.filter(i=>Pe.size(i.dims)>0);e.compute(Pm(a,o,n,t[0].dataType),{inputs:a})},Sm=e=>jt({axis:e.axis})}),_n,gn,wn,El,Mn=Ue(()=>{Mt(),Pt(),_n=(e,r,t="f32")=>{switch(e.activation){case"Relu":return`value = max(value, ${r}(0.0));`;case"Sigmoid":return`value = (${r}(1.0) / (${r}(1.0) + exp(-value)));`;case"Clip":return`value = clamp(value, ${r}(${t}(uniforms.clip_min)), ${r}(${t}(uniforms.clip_max)));`;case"HardSigmoid":return`value = max(${r}(0.0), min(${r}(1.0), ${t}(uniforms.alpha) * value + ${t}(uniforms.beta)));`;case"LeakyRelu":return`value = select(${t}(uniforms.alpha) * value, value, value >= ${r}(0.0));`;case"Tanh":return`let e2x = exp(-2.0 * abs(value));
              value = sign(value) * (1.0 - e2x) / (1.0 + e2x);
        `;case"":return"";default:throw new Error(`Unsupported activation ${e.activation}`)}},gn=(e,r)=>{e.activation==="Clip"?r.push({type:1,data:e.clipMax},{type:1,data:e.clipMin}):e.activation==="HardSigmoid"?r.push({type:1,data:e.alpha},{type:1,data:e.beta}):e.activation==="LeakyRelu"&&r.push({type:1,data:e.alpha})},wn=(e,r)=>{e.activation==="Clip"?r.push({name:"clip_max",type:"f32"},{name:"clip_min",type:"f32"}):e.activation==="HardSigmoid"?r.push({name:"alpha",type:"f32"},{name:"beta",type:"f32"}):e.activation==="LeakyRelu"&&r.push({name:"alpha",type:"f32"})},El=e=>{let r=(e==null?void 0:e.activation)||"";if(r==="HardSigmoid"){let[t,s]=(e==null?void 0:e.activation_params)||[.2,.5];return{activation:r,alpha:t,beta:s}}else if(r==="Clip"){let[t,s]=(e==null?void 0:e.activation_params)||[Jd,Yd];return{activation:r,clipMax:s,clipMin:t}}else if(r==="LeakyRelu"){let[t]=(e==null?void 0:e.activation_params)||[.01];return{activation:r,alpha:t}}return{activation:r}}}),zr,$m,Pl=Ue(()=>{zr=(e,r)=>{switch(e){case 1:return r;case 2:return`vec2<${r}>`;case 3:return`vec3<${r}>`;case 4:return`vec4<${r}>`;default:throw new Error(`${e}-component is not supported.`)}},$m=e=>`
      ${e?"value = value + getBiasByOutputCoords(coords);":""}
      `}),Am,tT=Ue(()=>{Am=e=>`
fn getIndexFromCoords4D(coords : vec4<i32>, shape : vec4<i32>) -> i32 {
  return dot(coords, vec4<i32>(
      shape.y * shape.z * shape.w, shape.z * shape.w, shape.w, 1));
}
fn getOutputIndexFromCoords(coords : vec4<i32>) -> i32 {
  return dot(coords, vec4<i32>(
    i32(${e}.x), i32(${e}.y), i32(${e}.z), 1));
}
`}),bo,Cl,Sl=Ue(()=>{Mt(),Pt(),Ct(),Mn(),bo=(e,r,t,s,o)=>{let n=s-t;return`
      ${Array.from({length:t}).map((a,i)=>`
      if (${it(r.shape,i,r.rank)} != 1) {
        ${r.indicesSet(e,i,it(o,i+n,s))}
      } else {
        ${r.indicesSet(e,i,0)}
      }`).join("")}
`},Cl=(e,r,t,s,o=!1,n)=>{let a=e[0].dims,i=e[1].dims,l=a[a.length-2],c=i[i.length-1],p=a[a.length-1],d=or(c),u=or(p),_=or(l),f=Pe.size(t)/d/_,b=e.length>2,A=s?s.slice(0,-2):t.slice(0,-2),g=[Pe.size(A),l,c],y=[{type:12,data:f},{type:12,data:l},{type:12,data:c},{type:12,data:p}];gn(r,y),y.push(...lt(A,a,i)),b&&y.push(...lt(e[2].dims)),y.push(...lt(g));let C=x=>{let M=ml("batch_dims",e[0].dataType,A.length),T=Oe("a",e[0].dataType,a.length,u),v=Oe("b",e[1].dataType,i.length,d),P=at("output",e[0].dataType,g.length,d),F=Sr(P.type.tensor),D=_n(r,P.type.value,F),K=[T,v],U="";if(b){let q=o?d:1;K.push(Oe("bias",e[2].dataType,e[2].dims.length,q)),U=`${o?`value += bias[col / ${q}];`:`value += ${P.type.value}(bias[row + i]);`}`}let j=[{name:"output_size",type:"u32"},{name:"M",type:"u32"},{name:"N",type:"u32"},{name:"K",type:"u32"}];wn(r,j);let ne=()=>{let q=`var a_data: ${T.type.value};`;for(let te=0;te<u;te++)q+=`
              let b_data${te} = b[(b_offset + (k + ${te}) * uniforms.N + col) / ${d}];`;for(let te=0;te<_;te++){q+=`a_data = a[(a_offset + (row + ${te}) * uniforms.K + k) / ${u}];`;for(let Z=0;Z<u;Z++)q+=`
            values[${te}] = fma(${v.type.value}(a_data${u===1?"":`[${Z}]`}), b_data${Z}, values[${te}]);
`}return q};return`
  ${x.registerUniforms(j).registerInternalVariables(M).declareVariables(...K,P)}
  ${x.mainStart()}
    ${x.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_size")}
    let col = (global_idx % (uniforms.N / ${d})) * ${d};
    var index1 = global_idx / (uniforms.N / ${d});
    let stride1 = uniforms.M / ${_};
    let row = (index1 % stride1) * ${_};
    let batch = index1 / stride1;

    ${t.length===2?"":`let batch_indices = ${M.offsetToIndices("batch")};`}

    var a_indices: ${T.type.indices};
    ${bo("a_indices",T,T.rank-2,M.rank,"batch_indices")}
    ${T.indicesSet("a_indices",T.rank-2,0)}
    ${T.indicesSet("a_indices",T.rank-1,0)}
    let a_offset = ${T.indicesToOffset("a_indices")};

    var b_indices: ${v.type.indices};
    ${bo("b_indices",v,v.rank-2,M.rank,"batch_indices")}
    ${v.indicesSet("b_indices",v.rank-2,0)}
    ${v.indicesSet("b_indices",v.rank-1,0)}
    let b_offset = ${v.indicesToOffset("b_indices")};
    var values: array<${P.type.value}, ${_}>;
    for (var k: u32 = 0u; k < uniforms.K; k = k + ${u}) {
      ${ne()}
    }
    for (var i = 0u; i < ${_}u; i++) {
      var value = values[i];
      ${U}
      ${D}
      let cur_indices = ${P.type.indices}(batch, row + i, col);
      let offset = ${P.indicesToOffset("cur_indices")};
      ${P.setByOffset(`offset / ${d}`,"value")};
    }
  }
  `};return{name:"MatMulNaive",shaderCache:{hint:`${r.activation};${d};${u};${_};${o}`,inputDependencies:b?["rank","rank","rank"]:["rank","rank"]},getRunData:()=>({outputs:[{dims:n?n(t):t,dataType:e[0].dataType}],dispatchGroup:{x:Math.ceil(f/64)},programUniforms:y}),getShaderSource:C}}}),km,Im,$l,Al,Fm,kl,Om,_a,Il=Ue(()=>{Mt(),Pt(),Ct(),Mn(),Sl(),Pl(),km=(e,r)=>e?`
        mm_Asub[inputRow][inputCol] = mm_readA(batch,
          kStart + inputRow,
          globalRowStart / innerElementSize + inputCol${r?", batchIndices":""});
        `:`
        mm_Asub[inputRow][inputCol] = mm_readA(batch,
          globalRow + innerRow,
          kStart / innerElementSize + inputCol${r?", batchIndices":""});
        `,Im=(e,r)=>e?`
        let ACached0 = mm_Asub[k * innerElementSize][localRow];
        let ACached1 = mm_Asub[k * innerElementSize + 1][localRow];
        let ACached2 = mm_Asub[k * innerElementSize + 2][localRow];
        ${r===3?"":"let ACached3 = mm_Asub[k * innerElementSize + 3][localRow];"}
        for (var i = 0; i < rowPerThread; i = i + 1) {
          acc[i] = BCached0 * ACached0[i] + acc[i];
          acc[i] = BCached1 * ACached1[i] + acc[i];
          acc[i] = BCached2 * ACached2[i] + acc[i];
          ${r===3?"":"acc[i] = BCached3 * ACached3[i] + acc[i];"}
        }`:`
        for (var i = 0; i < rowPerThread; i = i + 1) {
          let ACached = mm_Asub[tileRow + i][k];
          acc[i] = BCached0 * ACached.x + acc[i];
          acc[i] = BCached1 * ACached.y + acc[i];
          acc[i] = BCached2 * ACached.z + acc[i];
          ${r===3?"":"acc[i] = BCached3 * ACached.w + acc[i];"}
        }`,$l=(e,r,t="f32",s,o=!1,n=32,a=!1,i=32)=>{let l=r[1]*e[1],c=r[0]*e[0],p=o?l:n,d=o?n:l,u=p/r[0],_=n/r[1];if(!((o&&u===4&&e[1]===4||!o&&(u===3||u===4))&&p%r[0]===0&&n%r[1]===0&&e[0]===4))throw new Error(`If transposeA ${o} is true, innerElementSize ${u} and workPerThread[1] ${e[1]} must be 4.
      Otherwise, innerElementSize ${u} must be 3 or 4.
  tileAWidth ${p} must be divisible by workgroupSize[0]${r[0]}. tileInner ${n} must be divisible by workgroupSize[1] ${r[1]}. colPerThread ${e[0]} must be 4.`);return`
var<workgroup> mm_Asub: array<array<vec${u}<${t}>, ${p/u}>, ${d}>;
var<workgroup> mm_Bsub: array<array<vec4<${t}>, ${c/e[0]}>, ${n}>;

const rowPerThread = ${e[1]};
const colPerThread = ${e[0]};
const innerElementSize = ${u};
const tileInner = ${n};

@compute @workgroup_size(${r[0]}, ${r[1]}, ${r[2]})
fn main(@builtin(local_invocation_id) localId : vec3<u32>,
        @builtin(global_invocation_id) globalId : vec3<u32>,
        @builtin(workgroup_id) workgroupId : vec3<u32>) {
  let localRow = i32(localId.y);
  let tileRow = localRow * rowPerThread;
  let tileCol = i32(localId.x);

  let globalRow =i32(globalId.y) * rowPerThread;
  let globalCol = i32(globalId.x);
  let batch = ${a?"0":"i32(globalId.z)"};
  ${s?`let batchIndices = ${s.offsetToIndices("u32(batch)")};`:""}
  let globalRowStart = i32(workgroupId.y) * ${l};

  let num_tiles = ${a?`${Math.ceil(i/n)}`:"(uniforms.dim_inner - 1) / tileInner + 1"};
  var kStart = ${a?`i32(globalId.z) * ${i}`:"0"};

  var acc: array<vec4<${t}>, rowPerThread>;

  // Loop over shared dimension.
  let tileRowB = localRow * ${_};
  for (var t = 0; t < num_tiles; t = t + 1) {
      // Load one tile of A into local memory.
      for (var innerRow = 0; innerRow < rowPerThread; innerRow = innerRow + 1) {
          let inputRow = tileRow + innerRow;
          let inputCol = tileCol;
          ${km(o,s)}
      }

      // Load one tile of B into local memory.
      for (var innerRow = 0; innerRow < ${_}; innerRow = innerRow + 1) {
          let inputRow = tileRowB + innerRow;
          let inputCol = tileCol;
          mm_Bsub[inputRow][inputCol] = mm_readB(batch, kStart + inputRow, globalCol${s?", batchIndices":""});
      }
      kStart = kStart + tileInner;
      workgroupBarrier();

      // Compute acc values for a single thread.
      for (var k = 0; k < tileInner / innerElementSize; k = k + 1) {
          let BCached0 = mm_Bsub[k * innerElementSize][tileCol];
          let BCached1 = mm_Bsub[k * innerElementSize + 1][tileCol];
          let BCached2 = mm_Bsub[k * innerElementSize + 2][tileCol];
          ${u===3?"":"let BCached3 = mm_Bsub[k * innerElementSize + 3][tileCol];"}

          ${Im(o,u)}
      }

      workgroupBarrier();
  }

  for (var innerRow = 0; innerRow < rowPerThread; innerRow = innerRow + 1) {
      mm_write(batch, globalRow + innerRow, globalCol, acc[innerRow]);
  }
}`},Al=(e,r)=>e?`
            mm_Asub[inputRow][inputCol] = mm_readA(batch,
              kStart + inputRow,
              globalRowStart + inputCol${r?", batchIndices":""});
            `:`
            mm_Asub[inputRow][inputCol] = mm_readA(batch,
              globalRowStart + inputRow,
              kStart + inputCol${r?", batchIndices":""});
            `,Fm=e=>e?"let ACached = mm_Asub[k][tileRow + innerRow];":"let ACached = mm_Asub[tileRow + innerRow][k];",kl=(e,r,t="f32",s,o=!1,n=32,a=!1,i=32,l=!1)=>{let c=e[1]*r[1],p=e[0]*r[0],d=o?c:n,u=o?n:c;if(!(u%r[1]===0&&d%r[0]===0&&n%r[1]===0))throw new Error(`tileAHight ${u} must be divisible by workgroupSize[1]${r[1]}, tileAWidth ${d} must be divisible by workgroupSize[0]${r[0]}, tileInner ${n} must be divisible by workgroupSize[1]${r[1]}`);let _=u/r[1],f=d/r[0],b=n/r[1],A=l?`
    let localRow = i32(localId.y);
    let localCol = i32(localId.x);
    let globalRowStart = i32(workgroupId.y) * ${c};
    let globalColStart = i32(workgroupId.x) * ${p};

    // Loop over shared dimension.
    for (var t = 0; t < num_tiles; t = t + 1) {
      // Load one tile of A into local memory.
      for (var inputRow = localRow; inputRow < ${u}; inputRow = inputRow + ${r[1]}) {
        for (var inputCol = localCol; inputCol < ${d}; inputCol = inputCol + ${r[0]}) {
          ${Al(o,s)}
        }
      }
      // Load one tile of B into local memory.
      for (var inputRow = localRow; inputRow < ${n}; inputRow = inputRow + ${r[1]}) {
            for (var inputCol = localCol; inputCol < ${p}; inputCol = inputCol + ${r[0]}) {
          mm_Bsub[inputRow][inputCol] = mm_readB(batch,
            kStart + inputRow,
            globalColStart + inputCol${s?", batchIndices":""});
        }
      }
      kStart = kStart + tileInner;
      workgroupBarrier();

      // Compute acc values for a single thread.
      var BCached : array<${t}, colPerThread>;
      for (var k = 0; k < tileInner; k = k + 1) {
        for (var inner = 0; inner < colPerThread; inner = inner + 1) {
          BCached[inner] = mm_Bsub[k][localCol + inner * ${r[0]}];
        }
        for (var innerRow = 0; innerRow < rowPerThread; innerRow = innerRow + 1) {
          let ACached = ${o?`mm_Asub[k][localRow + innerRow * ${r[1]}];`:`mm_Asub[localRow + innerRow * ${r[1]}][k];`}
          for (var innerCol = 0; innerCol < colPerThread; innerCol = innerCol + 1) {
            acc[innerRow][innerCol] = acc[innerRow][innerCol] +
                ACached * BCached[innerCol];
          }
        }
      }
      workgroupBarrier();
    }
    for (var innerRow = 0; innerRow < rowPerThread; innerRow = innerRow + 1) {
      let gRow = globalRowStart + localRow + innerRow * ${r[1]};
      for (var innerCol = 0; innerCol < colPerThread; innerCol = innerCol + 1) {
        let gCol = globalColStart + localCol + innerCol * ${r[0]};
        mm_write(batch, gRow, gCol, acc[innerRow][innerCol]);
      }
    }
    `:`
let tileRow = i32(localId.y) * rowPerThread;
let tileCol = i32(localId.x) * colPerThread;

let globalRow = i32(globalId.y) * rowPerThread;
let globalCol = i32(globalId.x) * colPerThread;
let globalRowStart = i32(workgroupId.y) * ${c};

let tileRowA = i32(localId.y) * ${_};
let tileColA = i32(localId.x) * ${f};
let tileRowB = i32(localId.y) * ${b};
// Loop over shared dimension.
for (var t = 0; t < num_tiles; t = t + 1) {
  // Load one tile of A into local memory.
  for (var innerRow = 0; innerRow < ${_}; innerRow = innerRow + 1) {
    for (var innerCol = 0; innerCol < ${f}; innerCol = innerCol + 1) {
      let inputRow = tileRowA + innerRow;
      let inputCol = tileColA + innerCol;
      ${Al(o,s)}
    }
  }

  // Load one tile of B into local memory.
  for (var innerRow = 0; innerRow < ${b}; innerRow = innerRow + 1) {
    for (var innerCol = 0; innerCol < colPerThread; innerCol = innerCol + 1) {
      let inputRow = tileRowB + innerRow;
      let inputCol = tileCol + innerCol;
      mm_Bsub[inputRow][inputCol] = mm_readB(batch,
        kStart + inputRow,
        globalCol + innerCol${s?", batchIndices":""});
    }
  }
  kStart = kStart + tileInner;
  workgroupBarrier();

  // Compute acc values for a single thread.
  var BCached : array<${t}, colPerThread>;
  for (var k = 0; k < tileInner; k = k + 1) {
    for (var inner = 0; inner < colPerThread; inner = inner + 1) {
      BCached[inner] = mm_Bsub[k][tileCol + inner];
    }

    for (var innerRow = 0; innerRow < rowPerThread; innerRow = innerRow + 1) {
      ${Fm(o)}
      for (var innerCol = 0; innerCol < colPerThread; innerCol = innerCol + 1) {
        acc[innerRow][innerCol] = acc[innerRow][innerCol] + ACached * BCached[innerCol];
      }
    }
  }

  workgroupBarrier();
}

for (var innerRow = 0; innerRow < rowPerThread; innerRow = innerRow + 1) {
  for (var innerCol = 0; innerCol < colPerThread; innerCol = innerCol + 1) {
    mm_write(batch, globalRow + innerRow, globalCol + innerCol,
        acc[innerRow][innerCol]);
  }
}
`;return`
  var<workgroup> mm_Asub : array<array<${t}, ${d}>, ${u}>;
  var<workgroup> mm_Bsub : array<array<${t}, ${p}>, ${n}>;
  const rowPerThread = ${e[1]};
  const colPerThread = ${e[0]};
  const tileInner = ${n};

@compute @workgroup_size(${r[0]}, ${r[1]}, ${r[2]})
fn main(@builtin(local_invocation_id) localId : vec3<u32>,
        @builtin(global_invocation_id) globalId : vec3<u32>,
        @builtin(workgroup_id) workgroupId : vec3<u32>) {
    let batch = ${a?"0":"i32(globalId.z)"};
    ${s?`let batchIndices = ${s.offsetToIndices("u32(batch)")};`:""}
    let num_tiles = ${a?`${Math.ceil(i/n)}`:"(uniforms.dim_inner - 1) / tileInner + 1"};
    var kStart = ${a?`i32(globalId.z) * ${i}`:"0"};

    var acc : array<array<${t}, colPerThread>, rowPerThread>;
    ${A}
  }
`},Om=(e,r,t,s,o=!1)=>{let[n,a,i,l]=s,c=Sr(s[0].type.tensor);return`
    fn mm_readA(batch: i32, row: i32, colIn: i32, batchIndices: ${n.type.indices}) -> ${zr(e,c)} {
      var value = ${zr(e,c)}(0.0);
      let col = colIn * ${e};
      if(row < uniforms.dim_a_outer && col < uniforms.dim_inner)
      {
        var aIndices: ${a.type.indices};
        ${bo("aIndices",a,a.rank-2,n.rank,"batchIndices")}
        ${a.indicesSet("aIndices",a.rank-2,"u32(row)")}
        ${a.indicesSet("aIndices",a.rank-1,"u32(colIn)")}
        value = ${a.getByIndices("aIndices")};
      }
      return value;
    }

    fn mm_readB(batch: i32, row: i32, colIn: i32, batchIndices: ${n.type.indices}) -> ${zr(e,c)} {
      var value = ${zr(e,c)}(0.0);
      let col = colIn * ${e};
      if(row < uniforms.dim_inner && col < uniforms.dim_b_outer)
      {
        var bIndices: ${i.type.indices};
        ${bo("bIndices",i,i.rank-2,n.rank,"batchIndices")}
        ${i.indicesSet("bIndices",i.rank-2,"u32(row)")}
        ${i.indicesSet("bIndices",i.rank-1,"u32(colIn)")}
        value = ${i.getByIndices("bIndices")};
      }
      return value;
    }

    fn mm_write(batch: i32, row: i32, colIn: i32, valueIn: ${zr(e,c)}) {
      let col = colIn * ${e};
      if (row < uniforms.dim_a_outer && col < uniforms.dim_b_outer) {
        var value = valueIn;
        let coords = vec3<i32>(batch, row, colIn);
        ${r?`value = value + ${o?"bias[colIn]":`${zr(e,c)}(bias[row])`};`:""}
        ${t}
        ${l.setByIndices("vec3<u32>(coords)","value")}
      }
    }
    `},_a=(e,r,t,s,o=!1,n)=>{let a=e[0].dims,i=e[1].dims,l=a.slice(0,-2),c=i.slice(0,-2),p=s?s.slice(0,-2):t.slice(0,-2),d=Pe.size(p),u=a[a.length-2],_=a[a.length-1],f=i[i.length-1],b=_%4===0&&f%4===0,A=u<=8?[4,1,1]:[4,4,1],g=[8,8,1],y=[Math.ceil(f/g[0]/A[0]),Math.ceil(u/g[1]/A[1]),Math.ceil(d/g[2]/A[2])],C=b?4:1,x=[...l,u,_/C],M=x.length,T=[...c,_,f/C],v=T.length,P=[d,u,f/C],F=[{type:6,data:u},{type:6,data:f},{type:6,data:_}];gn(r,F),F.push(...lt(p,x,T));let D=["rank","rank"],K=e.length>2;K&&(F.push(...lt(e[2].dims)),D.push("rank")),F.push(...lt(P));let U=j=>{let ne=p.length,q=ml("batchDims",e[0].dataType,ne,1),te=Sr(e[0].dataType),Z=Oe("a",e[0].dataType,M,C),ae=Oe("b",e[1].dataType,v,C),he=at("result",e[0].dataType,P.length,C),Q=[Z,ae];if(K){let J=o?C:1;Q.push(Oe("bias",e[2].dataType,e[2].dims.length,J))}let B=[{name:"dim_a_outer",type:"i32"},{name:"dim_b_outer",type:"i32"},{name:"dim_inner",type:"i32"}];wn(r,B);let O=Sr(he.type.tensor),W=_n(r,he.type.value,O),N=Om(C,K,W,[q,Z,ae,he],o);return`
  ${j.registerUniforms(B).registerInternalVariables(q).declareVariables(...Q,he)}
  ${N}
  ${b?$l(A,g,te,q):kl(A,g,te,q)}
                   `};return{name:"MatMul",shaderCache:{hint:`${A};${r.activation};${b};${o}`,inputDependencies:D},getRunData:()=>({outputs:[{dims:n?n(t):t,dataType:e[0].dataType}],dispatchGroup:{x:y[0],y:y[1],z:y[2]},programUniforms:F}),getShaderSource:U}}}),Dm,Lm,rT=Ue(()=>{Mt(),js(),Ct(),Mn(),Pl(),tT(),Il(),Dm=(e,r,t,s,o=!1,n,a=4,i=4,l=4,c="f32")=>{let p=F=>{switch(F){case 1:return"resData = x[xIndex];";case 3:return`resData = vec3<${c}>(x[xIndex], x[xIndex + 1], x[xIndex + 2]);`;case 4:return"resData = x[xIndex / 4];";default:throw new Error(`innerElementSize ${F} is not supported.`)}},d=F=>{switch(F){case 1:return"return w[row * i32(uniforms.w_shape[3]) + colIn];";case 4:return"return w[row * i32(uniforms.w_shape[3]) / 4 + colIn];";default:throw new Error(`innerElementSize ${F} is not supported.`)}},u=e?`
    let coord = vec4<i32>(batch, xRow, xCol, xCh);
    `:`
    let coord = vec4<i32>(batch, xCh, xRow, xCol);
    `,_=e?`
    let coords = vec4<i32>(
      batch,
      row / outWidth,
      row % outWidth,
      col);
    `:`
    let coords = vec4<i32>(
      batch,
      row,
      col / outWidth,
      col % outWidth);
    `,f=e?"i32(uniforms.x_shape[1])":"i32(uniforms.x_shape[2])",b=e?"i32(uniforms.x_shape[2])":"i32(uniforms.x_shape[3])",A=e?"row":"col",g=e?"col":"row",y=`
    let inChannels = i32(uniforms.w_shape[2]);
    let outWidth = ${e?"i32(uniforms.result_shape[2])":"i32(uniforms.result_shape[3])"};
    let outRow = ${A} / outWidth;
    let outCol = ${A} % outWidth;

    let WRow = ${g} / (i32(uniforms.w_shape[1]) * inChannels);
    let WCol = ${g} / inChannels % i32(uniforms.w_shape[1]);
    let xRow = outRow * uniforms.stride[0] + uniforms.dilation[0] * WRow - uniforms.pad[0];
    let xCol = outCol * uniforms.stride[1] + uniforms.dilation[1] * WCol - uniforms.pad[1];
    let xCh = ${g} % inChannels;
    var resData = ${zr(a,c)}(0.0);
    // The bounds checking is always needed since we use it to pad zero for
    // the 'same' padding type.
    if (xRow >= 0 && xRow < ${f} && xCol >= 0 && xCol < ${b}) {
      ${u}
      let xIndex = getIndexFromCoords4D(coord, vec4<i32>(uniforms.x_shape));
      ${p(a)}
    }
    return resData;`,C=e?r&&s?`
    let col = colIn * ${a};
    ${y}`:`
    let col = colIn * ${a};
    if (row < uniforms.dim_a_outer && col < uniforms.dim_inner) {
      ${y}
    }
    return ${zr(a,c)}(0.0);`:s&&t?`
    let col = colIn * ${a};
    ${y}`:`
    let col = colIn * ${a};
    if (row < uniforms.dim_inner && col < uniforms.dim_b_outer) {
      ${y}
    }
    return ${zr(a,c)}(0.0);`,x=e?s&&t?d(i):`
    let col = colIn * ${i};
    if (row < uniforms.dim_inner && col < uniforms.dim_b_outer) {
      ${d(i)}
    }
    return ${zr(i,c)}(0.0);`:`
    let col = colIn * ${i};
    if (row < uniforms.dim_inner && col < uniforms.dim_a_outer) {
      ${d(i)}
    }
    return ${zr(i,c)}(0.0);`,M=zr(l,c),T=zr(e?a:i,c),v=zr(e?i:a,c),P=_n(n,M,c);return`
    fn mm_readA(batch: i32, row : i32, colIn : i32) -> ${T} {
      ${e?C:x}
    }

    fn mm_readB(batch: i32, row : i32, colIn : i32) -> ${v} {
      ${e?x:C}
    }

    fn mm_write(batch: i32, row : i32, colIn : i32, valueIn : ${M}) {
      let col = colIn * ${l};
      if (row < uniforms.dim_a_outer && col < uniforms.dim_b_outer)
      {
      var value = valueIn;
      let outWidth = ${e?"i32(uniforms.result_shape[2])":"i32(uniforms.result_shape[3])"};
      ${_}
      ${$m(o)}
      ${P}
      setOutputAtCoords(coords[0], coords[1], coords[2], coords[3], value);
      }
    }`},Lm=(e,r,t,s,o,n,a,i,l)=>{let c=r.format==="NHWC",p=c?e[0].dims[3]:e[0].dims[1],d=t[0],u=c?t[2]:t[3],_=c?t[1]:t[2],f=c?t[3]:t[1],b=c&&(p%4===0||p%3===0)&&f%4===0,A=c?f:u*_,g=c?u*_:f,y=[8,8,1],C=s<=8?[4,1,1]:[4,4,1],x=[Math.ceil(A/y[0]/C[0]),Math.ceil(g/y[1]/C[1]),Math.ceil(d/y[2]/C[2])];Dt("verbose",()=>`[conv2d_mm_webgpu] dispatch = ${x}`);let M=b?c&&p%4!==0?3:4:1,T=y[1]*C[1],v=y[0]*C[0],P=Math.max(y[0]*M,y[1]),F=s%T===0,D=o%v===0,K=n%P===0,U=b?[M,4,4]:[1,1,1],j=[{type:6,data:s},{type:6,data:o},{type:6,data:n},{type:6,data:[r.pads[0],r.pads[1]]},{type:6,data:r.strides},{type:6,data:r.dilations}];gn(r,j),j.push(...lt(e[0].dims,e[1].dims));let ne=["rank","rank"];a&&(j.push(...lt(e[2].dims)),ne.push("rank")),j.push(...lt(t));let q=te=>{let Z=[{name:"dim_a_outer",type:"i32"},{name:"dim_b_outer",type:"i32"},{name:"dim_inner",type:"i32"},{name:"pad",type:"i32",length:2},{name:"stride",type:"i32",length:2},{name:"dilation",type:"i32",length:2}];wn(r,Z);let ae=b?4:1,he=Sr(e[0].dataType),Q=`
      fn setOutputAtIndex(flatIndex : i32, value : ${b?`vec4<${he}>`:he}) {
        result[flatIndex] = ${b?`vec4<${he}>`:he}(value);
      }
      fn setOutputAtCoords(d0 : i32, d1 : i32, d2 : i32, d3 : i32, value : ${b?`vec4<${he}>`:he}) {
        let flatIndex = getOutputIndexFromCoords(vec4<i32>(d0, d1, d2, d3));
        setOutputAtIndex(flatIndex ${b?"/ 4":""}, value);
      }`,B=Oe("x",e[0].dataType,e[0].dims.length,M===3?1:M),O=Oe("w",e[1].dataType,e[1].dims.length,ae),W=[B,O],N=at("result",e[0].dataType,t.length,ae);if(a){let J=Oe("bias",e[2].dataType,e[2].dims.length,ae);W.push(J),Q+=`
        fn getBiasByOutputCoords(coords : vec4<i32>) -> ${b?`vec4<${he}>`:he} {
          return bias[coords.${c?"w":"y"}${b?"/ 4":""}];
        }`}return`
        ${Am("uniforms.result_strides")}
        //struct Uniforms { xShape : vec4<i32>, wShape : vec4<i32>, outShape : vec4<i32>,
        //  outShapeStrides: vec3<i32>, filterDims : vec2<i32>, pad : vec2<i32>, stride : vec2<i32>,
        //  dilation : vec2<i32>, dimAOuter : i32, dimBOuter : i32, dimInner : i32 };
        ${te.registerUniforms(Z).declareVariables(...W,N)}
        ${Q}
        ${Dm(c,F,D,K,a,r,U[0],U[1],U[2],he)}
        ${b?$l(C,y,he,void 0,!c,P):kl(C,y,he,void 0,!c,P,!1,void 0,i)}`};return{name:"Conv2DMatMul",shaderCache:{hint:`${r.cacheKey};${M};${b};${F};${D};${K};${T};${v};${P}`,inputDependencies:ne},getRunData:()=>({outputs:[{dims:l?l(t):t,dataType:e[0].dataType}],dispatchGroup:{x:x[0],y:x[1],z:x[2]},programUniforms:j}),getShaderSource:q}}}),zm,Fl,yo,Rm,Ol,Bm,Nm,jm,sT=Ue(()=>{Mt(),js(),Pt(),Ct(),Mn(),Pl(),zm=e=>{let r=1;for(let t=0;t<e.length;t++)r*=e[t];return r},Fl=e=>typeof e=="number"?[e,e,e]:e,yo=(e,r)=>r<=1?e:e+(e-1)*(r-1),Rm=(e,r,t,s=1)=>{let o=yo(r,s);return Math.floor((e[0]*(t-1)-t+o)/2)},Ol=(e,r,t,s,o)=>{o==null&&(o=Rm(e,r[0],s[0]));let n=[0,0,0,t];for(let a=0;a<3;a++)e[a]+2*o>=r[a]&&(n[a]=Math.trunc((e[a]-r[a]+2*o)/s[a]+1));return n},Bm=(e,r,t,s,o,n,a,i,l,c)=>{let p,d,u,_;if(e==="VALID"&&(e=0),typeof e=="number"){p={top:e,bottom:e,left:e,right:e,front:e,back:e};let f=Ol([r,t,s,1],[i,l,c],1,[o,n,a],e);d=f[0],u=f[1],_=f[2]}else if(Array.isArray(e)){if(!e.every((b,A,g)=>b===g[0]))throw Error(`Unsupported padding parameter: ${e}`);p={top:e[0],bottom:e[1],left:e[2],right:e[3],front:e[4],back:e[5]};let f=Ol([r,t,s,1],[i,l,c],1,[o,n,a],e[0]);d=f[0],u=f[1],_=f[2]}else if(e==="SAME_UPPER"){d=Math.ceil(r/o),u=Math.ceil(t/n),_=Math.ceil(s/a);let f=(d-1)*o+i-r,b=(u-1)*n+l-t,A=(_-1)*a+c-s,g=Math.floor(f/2),y=f-g,C=Math.floor(b/2),x=b-C,M=Math.floor(A/2),T=A-M;p={top:C,bottom:x,left:M,right:T,front:g,back:y}}else throw Error(`Unknown padding parameter: ${e}`);return{padInfo:p,outDepth:d,outHeight:u,outWidth:_}},Nm=(e,r,t,s,o,n=!1,a="channelsLast")=>{let i,l,c,p,d;if(a==="channelsLast")[i,l,c,p,d]=e;else if(a==="channelsFirst")[i,d,l,c,p]=e;else throw new Error(`Unknown dataFormat ${a}`);let[u,,_,f,b]=r,[A,g,y]=Fl(t),[C,x,M]=Fl(s),T=yo(_,C),v=yo(f,x),P=yo(b,M),{padInfo:F,outDepth:D,outHeight:K,outWidth:U}=Bm(o,l,c,p,A,g,y,T,v,P),j=n?u*d:u,ne=[0,0,0,0,0];return a==="channelsFirst"?ne=[i,j,D,K,U]:a==="channelsLast"&&(ne=[i,D,K,U,j]),{batchSize:i,dataFormat:a,inDepth:l,inHeight:c,inWidth:p,inChannels:d,outDepth:D,outHeight:K,outWidth:U,outChannels:j,padInfo:F,strideDepth:A,strideHeight:g,strideWidth:y,filterDepth:_,filterHeight:f,filterWidth:b,effectiveFilterDepth:T,effectiveFilterHeight:v,effectiveFilterWidth:P,dilationDepth:C,dilationHeight:x,dilationWidth:M,inShape:e,outShape:ne,filterShape:r}},jm=(e,r,t,s,o,n)=>{let a=n==="channelsLast";a?e[0].dims[3]:e[0].dims[1];let i=[64,1,1],l={x:t.map((A,g)=>g)},c=[Math.ceil(zm(l.x.map(A=>t[A]))/i[0]),1,1];Dt("verbose",()=>`[conv3d_naive_webgpu] dispatch = ${c}`);let p=1,d=Pe.size(t),u=[{type:12,data:d},{type:12,data:s},{type:12,data:o},{type:12,data:r.strides},{type:12,data:r.dilations}];gn(r,u),u.push(...lt(e[0].dims,e[1].dims));let _=["rank","rank"],f=e.length===3;f&&(u.push(...lt(e[2].dims)),_.push("rank")),u.push(...lt(t));let b=A=>{let g=[{name:"output_size",type:"u32"},{name:"filter_dims",type:"u32",length:s.length},{name:"pads",type:"u32",length:o.length},{name:"strides",type:"u32",length:r.strides.length},{name:"dilations",type:"u32",length:r.dilations.length}];wn(r,g);let y=1,C=Sr(e[0].dataType),x=Oe("x",e[0].dataType,e[0].dims.length,p),M=Oe("W",e[1].dataType,e[1].dims.length,y),T=[x,M],v=at("result",e[0].dataType,t.length,y),P="";if(f){let K=Oe("bias",e[2].dataType,e[2].dims.length,y);T.push(K),P+=`
        fn getBiasByOutputCoords(coords : array<u32, 5>) -> ${C} {
          return bias[${a?it("coords",4,5):it("coords",1,5)}];
        }`}let F=zr(p,C),D=_n(r,F,C);return`
            ${P}
            fn getX(d0 : u32, d1 : u32, d2 : u32, d3 : u32, d4 : u32) -> f32 {
              let aIndices = array<u32, 5>(d0, d1, d2, d3, d4);
              return ${x.getByIndices("aIndices")};
            }
            fn getW(d0 : u32, d1 : u32, d2 : u32, d3 : u32, d4 : u32) -> f32 {
              let aIndices = array<u32, 5>(d0, d1, d2, d3, d4);
              return ${M.getByIndices("aIndices")};
            }
          ${A.registerUniforms(g).declareVariables(...T,v)}
          ${A.mainStart()}
          ${A.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_size")}
              let coords = ${v.offsetToIndices("global_idx")};
              let batch = ${it("coords",0,x.rank)};
              let d2 = ${a?it("coords",x.rank-1,x.rank):it("coords",1,x.rank)};
              let xFRCCorner = vec3<u32>(${a?it("coords",1,x.rank):it("coords",2,x.rank)},
              ${a?it("coords",2,x.rank):it("coords",3,x.rank)},
              ${a?it("coords",3,x.rank):it("coords",4,x.rank)}) * uniforms.strides - uniforms.pads;
              let xFCorner = xFRCCorner.x;
              let xRCorner = xFRCCorner.y;
              let xCCorner = xFRCCorner.z;
              let xShapeY = ${a?it("uniforms.x_shape",1,x.rank):it("uniforms.x_shape",2,x.rank)};
              let xShapeZ = ${a?it("uniforms.x_shape",2,x.rank):it("uniforms.x_shape",3,x.rank)};
              let xShapeW = ${a?it("uniforms.x_shape",3,x.rank):it("uniforms.x_shape",4,x.rank)};
              let xShapeU = ${a?it("uniforms.x_shape",4,x.rank):it("uniforms.x_shape",1,x.rank)};
              let inputDepthNearestVec4 = (xShapeU / 4) * 4;
              let inputDepthVec4Remainder = xShapeU % 4;

              var value = 0.0;
              for (var wF = 0u; wF < uniforms.filter_dims[0]; wF++) {
                let xF = xFCorner + wF * uniforms.dilations[0];
                if (xF < 0 || xF >= xShapeY) {
                  continue;
                }

                for (var wR = 0u; wR < uniforms.filter_dims[1]; wR++) {
                  let xR = xRCorner + wR * uniforms.dilations[1];
                  if (xR < 0 || xR >= xShapeZ) {
                    continue;
                  }

                  for (var wC = 0u; wC < uniforms.filter_dims[2]; wC++) {
                    let xC = xCCorner + wC * uniforms.dilations[2];
                    if (xC < 0 || xC >= xShapeW) {
                      continue;
                    }

                    for (var d1 = 0u; d1 < inputDepthNearestVec4; d1 += 4) {
                      ${a?`let xValues = vec4<f32>(
                               getX(batch, xF, xR, xC, d1),
                               getX(batch, xF, xR, xC, d1 + 1),
                               getX(batch, xF, xR, xC, d1 + 2),
                               getX(batch, xF, xR, xC, d1 + 3));
                            `:`let xValues = vec4<f32>(
                               getX(batch, d1, xF, xR, xC),
                               getX(batch, d1 + 1, xF, xR, xC),
                               getX(batch, d1 + 2, xF, xR, xC),
                               getX(batch, d1 + 3, xF, xR, xC));
                            `}
                            let wValues = vec4<f32>(
                              getW(d2, d1, wF, wR, wC),
                              getW(d2, d1 + 1, wF, wR, wC),
                              getW(d2, d1 + 2, wF, wR, wC),
                              getW(d2, d1 + 3, wF, wR, wC));
                      value += dot(xValues, wValues);
                    }
                    if (inputDepthVec4Remainder == 1) {
                        ${a?`value += getX(batch, xF, xR, xC, inputDepthNearestVec4)
                          * getW(d2, inputDepthNearestVec4, wF, wR, wC);`:`value += getX(batch, inputDepthNearestVec4, xF, xR, xC)
                          * getW(d2, inputDepthNearestVec4, wF, wR, wC);`}
                    } else if (inputDepthVec4Remainder == 2) {
                      ${a?`let xValues = vec2<f32>(
                        getX(batch, xF, xR, xC, inputDepthNearestVec4),
                        getX(batch, xF, xR, xC, inputDepthNearestVec4 + 1));
                      `:`let xValues = vec2<f32>(
                        getX(batch, inputDepthNearestVec4, xF, xR, xC),
                        getX(batch, inputDepthNearestVec4 + 1, xF, xR, xC));
                    `}
                    let wValues = vec2<f32>(
                      getW(d2, inputDepthNearestVec4, wF, wR, wC),
                      getW(d2, inputDepthNearestVec4 + 1, wF, wR, wC));
                      value += dot(xValues, wValues);
                    } else if (inputDepthVec4Remainder == 3) {
                      ${a?`let xValues = vec3<f32>(
                        getX(batch, xF, xR, xC, inputDepthNearestVec4),
                        getX(batch, xF, xR, xC, inputDepthNearestVec4 + 1),
                        getX(batch, xF, xR, xC, inputDepthNearestVec4 + 2));
                      `:`let xValues = vec3<f32>(
                        getX(batch, inputDepthNearestVec4, xF, xR, xC),
                        getX(batch, inputDepthNearestVec4 + 1, xF, xR, xC),
                        getX(batch, inputDepthNearestVec4 + 2, xF, xR, xC));
                    `}
                    let wValues = vec3<f32>(
                      getW(d2, inputDepthNearestVec4, wF, wR, wC),
                      getW(d2, inputDepthNearestVec4 + 1, wF, wR, wC),
                      getW(d2, inputDepthNearestVec4 + 2, wF, wR, wC));
                      value += dot(xValues, wValues);
                    }
                  }
                }
              }
              ${f?"value = value + getBiasByOutputCoords(coords)":""};
              ${D}
              result[global_idx] = f32(value);
          }`};return{name:"Conv3DNaive",shaderCache:{hint:`${r.cacheKey};${a};${p};${f}`,inputDependencies:_},getRunData:()=>({outputs:[{dims:t,dataType:e[0].dataType}],dispatchGroup:{x:c[0],y:c[1],z:c[2]},programUniforms:u}),getShaderSource:b}}}),Vm,Um,nT=Ue(()=>{Mt(),Pt(),Ct(),Mn(),Vm=(e,r,t,s)=>{let o=e.length>2,n=o?"value += b[output_channel];":"",a=e[0].dims,i=e[1].dims,l=r.format==="NHWC",c=l?t[3]:t[1],p=c/r.group,d=l&&p>=4?or(c):1,u=Pe.size(t)/d,_=[{type:12,data:u},{type:12,data:r.dilations},{type:12,data:[r.strides[0],r.strides[1]]},{type:12,data:[r.pads[0],r.pads[1]]},{type:12,data:p}];gn(r,_),_.push(...lt(a,[i[0],i[1],i[2],i[3]/d]));let f=o?["rank","rank","rank"]:["rank","rank"];_.push(...lt([t[0],t[1],t[2],t[3]/d]));let b=A=>{let g=at("output",e[0].dataType,t.length,d),y=Sr(g.type.tensor),C=_n(r,g.type.value,y),x=Oe("x",e[0].dataType,a.length),M=Oe("w",e[1].dataType,i.length,d),T=[x,M];o&&T.push(Oe("b",e[2].dataType,e[2].dims,d));let v=[{name:"output_size",type:"u32"},{name:"dilations",type:"u32",length:r.dilations.length},{name:"strides",type:"u32",length:2},{name:"pads",type:"u32",length:2},{name:"output_channels_per_group",type:"u32"}];wn(r,v);let P=l?`
      for (var wHeight: u32 = 0u; wHeight < uniforms.w_shape[0]; wHeight++) {
        let xHeight = xRCCorner.x + wHeight * uniforms.dilations[0];

        if (xHeight < 0u || xHeight >= uniforms.x_shape[1]) {
          continue;
        }

        for (var wWidth: u32 = 0u; wWidth < uniforms.w_shape[1]; wWidth++) {
          let xWidth = xRCCorner.y + wWidth * uniforms.dilations[1];
          if (xWidth < 0u || xWidth >= uniforms.x_shape[2]) {
            continue;
          }

          for (var wInChannel: u32 = 0u; wInChannel < uniforms.w_shape[2]; wInChannel++) {
            let input_channel = in_channel_offset + wInChannel;
            let xVal = ${x.get("batch","xHeight","xWidth","input_channel")};
            let wVal = ${M.get("wHeight","wWidth","wInChannel","output_channel")};
            value += xVal * wVal;
          }
        }
      }
      `:`
      for (var wInChannel: u32 = 0u; wInChannel < uniforms.w_shape[1]; wInChannel++) {
        let input_channel = in_channel_offset + wInChannel;
        for (var wHeight: u32 = 0u; wHeight < uniforms.w_shape[2]; wHeight++) {
          let xHeight = xRCCorner.x + wHeight * uniforms.dilations[0];

          if (xHeight < 0u || xHeight >= uniforms.x_shape[2]) {
            continue;
          }

          for (var wWidth: u32 = 0u; wWidth < uniforms.w_shape[3]; wWidth++) {
            let xWidth = xRCCorner.y + wWidth * uniforms.dilations[1];
            if (xWidth < 0u || xWidth >= uniforms.x_shape[3]) {
              continue;
            }

            let xVal = ${x.get("batch","input_channel","xHeight","xWidth")};
            let wVal = ${M.get("output_channel","wInChannel","wHeight","wWidth")};
            value += xVal * wVal;
          }
        }
      }
      `;return`
  ${A.registerUniforms(v).declareVariables(...T,g)}

  ${A.mainStart()}
    ${A.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_size")}

    let outputIndices = ${g.offsetToIndices("global_idx")};
    let batch: u32 = outputIndices[0];
    let output_channel: u32 = outputIndices[${l?3:1}];
    let xRCCorner: vec2<u32> = vec2<u32>(outputIndices[${l?1:2}], outputIndices[${l?2:3}]) * uniforms.strides - uniforms.pads;
    let group_id: u32 = output_channel * ${d} / uniforms.output_channels_per_group;
    var in_channel_offset = group_id * uniforms.w_shape[${l?2:1}];

    var value: ${g.type.value} = ${g.type.value}(0);
    ${P}
    ${n}
    ${C}
    ${g.setByOffset("global_idx","value")}
  }`};return{name:"GroupedConv",shaderCache:{hint:`${r.cacheKey}_${d}`,inputDependencies:f},getRunData:()=>({outputs:[{dims:s?s(t):t,dataType:e[0].dataType}],dispatchGroup:{x:Math.ceil(u/64)},programUniforms:_}),getShaderSource:b}},Um=(e,r,t,s)=>{let o=e.length>2,n=or(t[3]),a=or(t[2]),i=Pe.size(t)/n/a,l=[e[0].dims[0],e[0].dims[1],e[0].dims[2],e[0].dims[3]/n],c=[e[1].dims[0],e[1].dims[1],e[1].dims[2],e[1].dims[3]/n],p=[t[0],t[1],t[2],t[3]/n],d=[{type:12,data:i},{type:6,data:[r.strides[0],r.strides[1]]},{type:6,data:[r.pads[0],r.pads[1]]}];gn(r,d),d.push(...lt(l,c,p));let u=(a-1)*r.strides[1]+c[1],_=f=>{let b=at("output",e[0].dataType,p.length,n),A=Sr(b.type.tensor),g=_n(r,b.type.value,A),y=Oe("x",e[0].dataType,l.length,n),C=Oe("w",e[1].dataType,c.length,n),x=[y,C];o&&x.push(Oe("b",e[2].dataType,e[2].dims,n));let M=o?"value += b[output_channel];":"",T=[{name:"output_size",type:"u32"},{name:"strides",type:"i32",length:2},{name:"pads",type:"i32",length:2}];return wn(r,T),`
  ${f.registerUniforms(T).declareVariables(...x,b)}
  ${f.mainStart()}
    ${f.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_size")}
    let width0 = uniforms.output_shape[3];
    let output_channel = global_idx % width0;
    var index1 = global_idx / width0;
    let width1 = uniforms.output_shape[2] / ${a}u;
    let col = (index1 % width1) * ${a}u;
    index1 = index1 / width1;
    let row = index1 % uniforms.output_shape[1];
    let batch = index1 / uniforms.output_shape[1];

    let x_corner = vec2<i32>(i32(row), i32(col)) * uniforms.strides - uniforms.pads;

    var x_vals: array<${y.type.value}, ${u}>;
    var values: array<${b.type.value}, ${a}>;
    let input_channel = output_channel;
    // Use constant instead of uniform can give better performance for w's height/width.
    for (var w_height: u32 = 0u; w_height < ${c[0]}; w_height++) {
      let x_height = x_corner.x + i32(w_height);
      if (x_height >= 0 && u32(x_height) < uniforms.x_shape[1]) {
        for (var i = 0; i < ${u}; i++) {
          let x_width = x_corner.y + i;
          if (x_width >= 0 && u32(x_width) < uniforms.x_shape[2]) {
            x_vals[i] = ${y.get("batch","u32(x_height)","u32(x_width)","input_channel")};
          } else {
            x_vals[i] = ${y.type.value}(0);
          }
        }
        for (var w_width: u32 = 0u; w_width < ${c[1]}; w_width++) {
          let w_val = ${C.get("w_height","w_width","0","output_channel")};
          for (var i = 0u; i < ${a}u; i++) {
            values[i] = fma(x_vals[i * u32(uniforms.strides[1]) + w_width], w_val, values[i]);
          }
        }
      }
    }

    for (var i = 0u; i < ${a}u; i++) {
      var value = values[i];
      ${M}
      ${g}
      ${b.set("batch","row","col + i","output_channel","value")};
    }
  }`};return{name:"GroupedConv-Vectorize",shaderCache:{hint:`${r.cacheKey};${n};${a};${u};${c[0]};${c[1]}`,inputDependencies:o?["rank","rank","type"]:["rank","rank"]},getRunData:()=>({outputs:[{dims:s?s(t):t,dataType:e[0].dataType}],dispatchGroup:{x:Math.ceil(i/64)},programUniforms:d}),getShaderSource:_}}}),Wm,ga,Gm,wa,Dl,Ll,Hm,Km,zl,oT=Ue(()=>{Pt(),rT(),sT(),Il(),nT(),Mn(),Sl(),Ys(),Wm=(e,r,t,s,o,n)=>{let a=e[0],i=e.slice(n?1:2,n?3:4),l=i.length,c=r[0],p=r.slice(2).map((u,_)=>u+(u-1)*(t[_]-1)),d=i.map((u,_)=>u+s[_]+s[_+l]).map((u,_)=>Math.floor((u-p[_]+o[_])/o[_]));return d.splice(0,0,a),d.splice(n?3:1,0,c),d},ga=[2,3,1,0],Gm=(e,r)=>{if(!e||e.length!==2&&e.length!==3)throw new Error("Conv requires 2 or 3 inputs");if(e[0].dims.length>5)throw new Error("greater than 5D is not supported");if(e[0].dims.length!==e[1].dims.length)throw new Error("filter does not have same dimension as input");let t=e[0].dims[r.format==="NHWC"?e[0].dims.length-1:1],s=e[1].dims[1]*r.group;if(t!==s)throw new Error("FILTER_IN_CHANNEL should be equal to DATA_CHANNEL");if(e.length===3&&(e[2].dims.length!==1||e[1].dims[0]!==e[2].dims[0]))throw new Error("invalid bias");let o=e[0].dims.length-2;if(r.dilations.length!==o)throw new Error(`dilations should be ${o}D`);if(r.strides.length!==o)throw new Error(`strides should be ${o}D`);if(r.pads.length!==o*2)throw new Error(`pads should be ${o*2}D`);if(r.kernelShape.length!==0&&r.kernelShape.length!==e[1].dims.length-2)throw new Error("invalid kernel shape")},wa=(e,r)=>{let t=e.kernelShape.slice();t.length<r[1].dims.length-2&&t.push(...Array(r[1].dims.length-2-t.length).fill(0));for(let n=2;n<r[1].dims.length;++n)t[n-2]===0&&(t[n-2]=r[1].dims[n]);let s=e.pads.slice();la.adjustPadsBasedOnAutoPad(r[0].dims,e.strides,e.dilations,t,s,e.format==="NHWC",e.autoPad);let o=Object.assign({},e);return Object.assign(o,{kernelShape:t,pads:s}),o},Dl=e=>{let r=El(e),t=e.format,s=["NOTSET","VALID","SAME_UPPER","SAME_LOWER"][e.auto_pad],o=e.dilations,n=e.group,a=e.kernel_shape,i=e.pads,l=e.strides,c=e.w_is_const();return{autoPad:s,format:t,dilations:o,group:n,kernelShape:a,pads:i,strides:l,wIsConst:c,...r,cacheKey:`${e.format};${r.activation};`}},Ll=(e,r,t,s)=>{let o=t.format==="NHWC",n=Wm(r[0].dims,r[1].dims,t.dilations,t.pads,t.strides,o);if(t.group!==1){let T=[r[0]];if(o){let v=e.kernelCustomData.wT??e.compute(rs(r[1],ga),{inputs:[1],outputs:[t.wIsConst?-2:-1]})[0];t.wIsConst&&!e.kernelCustomData.wT&&(e.kernelCustomData.wT=v),T.push(v)}else T.push(r[1]);r.length===3&&T.push(r[2]),!e.adapterInfo.isArchitecture("ampere")&&o&&r[1].dims[0]===t.group&&r[1].dims[1]===1&&t.dilations[0]===1&&t.dilations[1]===1?e.compute(Um(T,t,n,s),{inputs:T}):e.compute(Vm(T,t,n,s),{inputs:T});return}let a=r.length===3,i=r[0].dims[o?1:2],l=r[0].dims[o?2:3],c=r[0].dims[o?3:1],p=r[1].dims[2],d=r[1].dims[3],u=n[o?1:2],_=n[o?2:3],f=n[o?3:1],b=o&&p===i&&d===l&&t.pads[0]===0&&t.pads[1]===0;if(b||p===1&&d===1&&t.dilations[0]===1&&t.dilations[1]===1&&t.strides[0]===1&&t.strides[1]===1&&t.pads[0]===0&&t.pads[1]===0){let T=n[0],v,P,F,D=[];if(o){let j=e.kernelCustomData.wT??e.compute(rs(r[1],ga),{inputs:[1],outputs:[t.wIsConst?-2:-1]})[0];if(t.wIsConst&&!e.kernelCustomData.wT&&(e.kernelCustomData.wT=j),b){let ne=i*l*c;v=r[0].reshape([1,T,ne]),P=j.reshape([1,ne,f]),F=[1,T,f]}else v=r[0].reshape([T,i*l,c]),P=j.reshape([1,c,f]),F=[T,u*_,f];D.push(v),D.push(P)}else v=r[0].reshape([T,c,i*l]),P=r[1].reshape([1,f,c]),F=[T,f,u*_],D.push(P),D.push(v);a&&D.push(r[2]);let K=F[2],U=D[0].dims[D[0].dims.length-1];K<8&&U<8?e.compute(Cl(D,t,n,F,o,s),{inputs:D}):e.compute(_a(D,t,n,F,o,s),{inputs:D});return}let A=!0,g=e.kernelCustomData.wT??e.compute(rs(r[1],ga),{inputs:[1],outputs:[t.wIsConst?-2:-1]})[0];t.wIsConst&&!e.kernelCustomData.wT&&(e.kernelCustomData.wT=g);let y=[r[0],g];a&&y.push(r[2]);let C=o?u*_:f,x=o?f:u*_,M=p*d*c;e.compute(Lm(y,t,n,C,x,M,a,A,s),{inputs:y})},Hm=(e,r)=>{let t=r.format==="NHWC",s=[e.inputs[0].reshape(t?[e.inputs[0].dims[0],1,e.inputs[0].dims[1],e.inputs[0].dims[2]]:[e.inputs[0].dims[0],e.inputs[0].dims[1],1,e.inputs[0].dims[2]]),e.inputs[1].reshape([e.inputs[1].dims[0],e.inputs[1].dims[1],1,e.inputs[1].dims[2]])];e.inputs.length===3&&s.push(e.inputs[2]);let o=[0,r.pads[0],0,r.pads[1]],n=[1].concat(r.strides),a=[1].concat(r.dilations),i=[1].concat(r.kernelShape),l=wa({...r,pads:o,strides:n,dilations:a,kernelShape:i},s);Ll(e,s,l,c=>t?[c[0],c[2],c[3]]:[c[0],c[1],c[3]])},Km=(e,r,t)=>{let s=t.format==="NHWC"?"channelsLast":"channelsFirst",o=wa(t,r),n=t.autoPad==="NOTSET"?t.pads:t.autoPad,a=Nm(r[0].dims,r[1].dims,t.strides,t.dilations,n,!1,s);e.compute(jm(r,o,a.outShape,[a.filterDepth,a.filterHeight,a.filterWidth],[a.padInfo.front,a.padInfo.top,a.padInfo.left],s))},zl=(e,r)=>{if(Gm(e.inputs,r),e.inputs[0].dims.length===3)Hm(e,r);else if(e.inputs[0].dims.length===5)Km(e,e.inputs,r);else{let t=wa(r,e.inputs);Ll(e,e.inputs,t)}}}),qm,aT=Ue(()=>{Mt(),js(),Pt(),Ct(),qm=(e,r,t)=>{let s=e.length>2,o=r.outputShape,n=r.format==="NHWC",a=r.group,i=e[1].dims,l=i[2]/a,c=i[3],p=n?or(l):1,d=n&&c===1&&l>=4,u=d?Math.floor(l/4)*4:Math.floor(l/p)*p,_=l-u,f=n?or(c):1,b=n?c===1?p:f:1,A=Pe.size(o)/f,g=[Math.ceil(A/64),1,1];Dt("verbose",()=>`[conv2d_backprop_webgpu] dispatch = ${g}`);let y=["rank","rank"],C=[r.strides[0],r.strides[1]],x=[r.kernelShape[n?1:2],r.kernelShape[n?2:3]],M=[r.dilations[0],r.dilations[1]],T=[x[0]+(r.dilations[0]<=1?0:(r.kernelShape[n?1:2]-1)*(r.dilations[0]-1)),x[1]+(r.dilations[1]<=1?0:(r.kernelShape[n?2:3]-1)*(r.dilations[1]-1))],v=[T[0]-1-Math.floor((r.pads[0]+r.pads[2])/2),T[1]-1-Math.floor((r.pads[1]+r.pads[3])/2)],P=[{type:12,data:A},{type:12,data:C},{type:12,data:x},{type:12,data:M},{type:12,data:T},{type:6,data:v},{type:12,data:u},{type:12,data:l},{type:12,data:c},...lt(e[0].dims,e[1].dims)];s&&(P.push(...lt(e[2].dims)),y.push("rank")),P.push(...lt(o));let F=D=>{let K=[{name:"output_size",type:"u32"},{name:"strides",type:"u32",length:C.length},{name:"filter_dims",type:"u32",length:x.length},{name:"dilations",type:"u32",length:x.length},{name:"effective_filter_dims",type:"u32",length:T.length},{name:"pads",type:"i32",length:v.length},{name:"input_channels_per_group_int",type:"u32"},{name:"input_channels_per_group",type:"u32"},{name:"output_channels_per_group",type:"u32"}],U=Sr(e[0].dataType),j=n?1:2,ne=n?2:3,q=n?3:1,te=Oe("W",e[1].dataType,e[1].dims.length,b),Z=Oe("Dy",e[0].dataType,e[0].dims.length,p),ae=[Z,te];s&&ae.push(Oe("bias",e[2].dataType,[o[q]].length,f));let he=at("result",e[0].dataType,o.length,f),Q=()=>{let W="";if(d)p===4?W+=`
        let xValue = ${Z.getByOffset("x_offset")};
        let wValue = ${te.getByOffset("w_offset")};
        dotProd = dotProd + dot(xValue, wValue);
        x_offset += 1u;
        w_offset += 1u;`:p===2?W+=`
          dotProd = dotProd + dot(vec4<${U}>(${Z.getByOffset("x_offset")}, ${Z.getByOffset("x_offset + 1u")}), vec4<${U}>(${te.getByOffset("w_offset")}, ${te.getByOffset("w_offset + 1u")}));
          x_offset += 2u;
          w_offset += 2u;`:p===1&&(W+=`
          dotProd = dotProd + dot(vec4<${U}>(${Z.getByOffset("x_offset")}, ${Z.getByOffset("x_offset + 1u")}, ${Z.getByOffset("x_offset + 2u")}, ${Z.getByOffset("x_offset + 3u")}), vec4<${U}>(${te.getByOffset("w_offset")}, ${te.getByOffset("w_offset + 1u")}, ${te.getByOffset("w_offset + 2u")}, ${te.getByOffset("w_offset + 3u")}));
          x_offset += 4u;
          w_offset += 4u;`);else if(W+=`
                  let xValue = ${n?Z.getByOffset(`${Z.indicesToOffset(`${Z.type.indices}(batch, idyR, idyC, inputChannel)`)} / ${p}`):Z.get("batch","inputChannel","idyR","idyC")};
        `,p===1)W+=`
          let w_offset = ${te.indicesToOffset(`${te.type.indices}(u32(wRPerm), u32(wCPerm), inputChannel, wOutChannel)`)};
          let wValue = ${te.getByOffset(`w_offset / ${b}`)};
          dotProd = dotProd + xValue * wValue;`;else for(let N=0;N<p;N++)W+=`
            let wValue${N} = ${te.getByOffset(`${te.indicesToOffset(`${te.type.indices}(u32(wRPerm), u32(wCPerm), inputChannel + ${N}, wOutChannel)`)} / ${b}`)};
            dotProd = dotProd + xValue[${N}] * wValue${N};`;return W},B=()=>{if(_===0)return"";if(!d)throw new Error(`packInputAs4 ${d} is not true.`);let W="";if(p===1){W+="dotProd = dotProd";for(let N=0;N<_;N++)W+=`
            + ${Z.getByOffset(`x_offset + ${N}`)} * ${te.getByOffset(`w_offset + ${N}`)}`;W+=";"}else if(p===2){if(_!==2)throw new Error(`Invalid inputChannelsRemainder ${_}.`);W+=`
          let xValue = ${Z.getByOffset("x_offset")};
          let wValue = ${te.getByOffset("w_offset")};
          dotProd = dotProd + dot(xValue, wValue);`}return W},O=`
            let outputIndices = ${he.offsetToIndices(`global_idx * ${f}`)};
            let batch = ${he.indicesGet("outputIndices",0)};
            let d1 = ${he.indicesGet("outputIndices",q)};
            let r = ${he.indicesGet("outputIndices",j)};
            let c = ${he.indicesGet("outputIndices",ne)};
            let dyCorner = vec2<i32>(i32(r), i32(c)) - uniforms.pads;
            let dyRCorner = dyCorner.x;
            let dyCCorner = dyCorner.y;
            let groupId = d1 / uniforms.output_channels_per_group;
            let wOutChannel = d1 - groupId * uniforms.output_channels_per_group;
            // Convolve dy(?, ?, d2) with w(:, :, d1, d2) to compute dx(xR, xC, d1).
            // ? = to be determined. : = across all values in that axis.
            var dotProd = ${he.type.value}(0.0);
            var wR: u32 = 0;
            if (uniforms.dilations.x == 1) {
              // Minimum wR >= 0 that satisfies (dyRCorner + wR) % (uniforms.strides.x) == 0
              wR = u32(((dyRCorner + i32(uniforms.strides.x) - 1) / i32(uniforms.strides.x)) * i32(uniforms.strides.x) - dyRCorner);
            }
            for (; wR < uniforms.effective_filter_dims.x; wR = wR + 1) {
              if (wR % uniforms.dilations.x != 0) {
                continue;
              }
              let dyR = (${U}(dyRCorner) + ${U}(wR)) / ${U}(uniforms.strides[0]);
              let wRPerm = uniforms.filter_dims.x - 1 - wR / uniforms.dilations.x;
              if (dyR < 0.0 || dyR >= ${U}(uniforms.Dy_shape[${j}]) || fract(dyR) > 0.0 ||
                  wRPerm < 0) {
                continue;
              }
              let idyR: u32 = u32(dyR);
              var wC: u32 = 0;
              if (uniforms.dilations.y == 1) {
                // Minimum wC >= 0 that satisfies (dyCCorner + wC) % (uniforms.strides.y) == 0
                wC = u32(((dyCCorner + i32(uniforms.strides.y) - 1) / i32(uniforms.strides.y)) * i32(uniforms.strides.y) - dyCCorner);
              }
              for (; wC < uniforms.effective_filter_dims.y; wC = wC + 1) {
                if (wC % uniforms.dilations.y != 0) {
                  continue;
                }
                let dyC = (${U}(dyCCorner) + ${U}(wC)) / ${U}(uniforms.strides.y);
                let wCPerm = uniforms.filter_dims.y - 1 - wC / uniforms.dilations.y;
                if (dyC < 0.0 || dyC >= ${U}(uniforms.Dy_shape[${ne}]) ||
                    fract(dyC) > 0.0 || wCPerm < 0) {
                  continue;
                }
                let idyC: u32 = u32(dyC);
                var inputChannel = groupId * uniforms.input_channels_per_group;
                ${d?`
                var x_offset = ${Z.indicesToOffset(`${Z.type.indices}(batch, idyR, idyC, inputChannel)`)} / ${p};
                var w_offset = ${te.indicesToOffset(`${te.type.indices}(wRPerm, wCPerm, inputChannel, wOutChannel)`)} / ${b};
                  `:""}
                for (var d2: u32 = 0; d2 < uniforms.input_channels_per_group_int; d2 = d2 + ${d?4:p}) {
                  ${Q()}
                  inputChannel = inputChannel + ${d?4:p};
                }
                ${B()}
                wC = wC + uniforms.strides.y - 1;
              }
              wR = wR + uniforms.strides[0] - 1;
            }
            let value = dotProd${s?` + bias[d1 / ${f}]`:""};
            ${he.setByOffset("global_idx","value")};
          `;return`
    ${D.registerUniforms(K).declareVariables(...ae,he)}
      ${D.mainStart()}
      ${D.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_size")};
    ${O}}`};return{name:"ConvTranspose2D",shaderCache:{hint:`${r.cacheKey};${p}${b}${f}${d}${_}`,inputDependencies:y},getRunData:()=>({dispatchGroup:{x:g[0],y:g[1],z:g[2]},outputs:[{dims:t?t(o):o,dataType:e[0].dataType}],programUniforms:P}),getShaderSource:F}}}),Qm,Xm,Jm,Rl,Ym,Zm,Bl,ef,tf,iT=Ue(()=>{aT(),Mn(),Ys(),Qm=(e,r,t,s,o,n)=>(e-1)*r+t+(s-1)*o+1-n,Xm=(e,r,t,s,o)=>{let n=Math.floor(e/2);r==="SAME_UPPER"?(t[s]=n,t[o]=e-n):r==="SAME_LOWER"&&(t[s]=e-n,t[o]=n)},Jm=(e,r,t,s,o,n,a,i,l,c)=>{let p=e.length-2,d=c.length===0;l.length<p&&l.push(...Array(p-l.length).fill(0));let u=e[0],_=r[i?3:1]*o;for(let f=0,b=e.length-p-(i?1:0);f<p;++f,++b){let A=e[b],g=d?A*a[f]:c[f],y=Qm(A,a[f],n[f],r[b],t[f],g);Xm(y,s,n,f,f+p),d&&c.push(a[f]*(A-1)+l[f]+(r[b]-1)*t[f]+1-n[f]-n[f+p])}c.splice(0,0,u),c.splice(i?3:1,0,_)},Rl=(e,r)=>{let t=e.kernelShape.slice();if(e.kernelShape.length===0||e.kernelShape.reduce((d,u)=>d*u,1)===0){t.length=0;for(let d=2;d<r[1].dims.length;++d)t.push(r[1].dims[d])}let s=e.format==="NHWC";t.splice(0,0,r[1].dims[0]),t.splice(s?3:1,0,r[1].dims[1]);let o=e.pads.slice(),n=e.outputShape.slice(),a=e.outputPadding.slice(),i=r[0].dims,l=e.dilations.slice();if(l.reduce((d,u)=>d+u,0)===0){let d=r[0].dims.length-2;l=new Array(d).fill(1)}let c=e.strides.slice();if(c.reduce((d,u)=>d+u,0)===0){let d=r[0].dims.length-2;c=new Array(d).fill(1)}Jm(i,t,l,e.autoPad,e.group,o,c,s,a,n);let p=Object.assign({},e);return Object.assign(p,{kernelShape:t,pads:o,outputPadding:a,outputShape:n,dilations:l,strides:c}),p},Ym=e=>{let r=El(e),t=e.format,s=["NOTSET","VALID","SAME_UPPER","SAME_LOWER"][typeof e.autoPad>"u"?0:e.autoPad],o=e.dilations,n=e.group,a=e.kernelShape,i=e.pads,l=e.strides,c=e.wIsConst(),p=e.outputPadding,d=e.outputShape;return{autoPad:s,format:t,dilations:o,group:n,kernelShape:a,outputPadding:p,outputShape:d,pads:i,strides:l,wIsConst:c,...r,cacheKey:`${e.format};${r.activation};`}},Zm=(e,r)=>{if(!e||e.length!==2&&e.length!==3)throw new Error("Conv requires 2 or 3 inputs");if(e[0].dims.length!==4&&e[0].dims.length!==3)throw new Error("currently only support 2-dimensional conv");if(e[0].dims.length!==e[1].dims.length)throw new Error("filter does not have same dimension as input");let t=e[0].dims[r.format==="NHWC"?e[0].dims.length-1:1],s=e[1].dims[0];if(t!==s)throw new Error("FILTER_IN_CHANNEL should be equal to DATA_CHANNEL");let o=e[1].dims[1]*r.group;if(e.length===3&&(e[2].dims.length!==1||e[2].dims[0]!==o))throw new Error("invalid bias");let n=e[0].dims.length-2;if(r.dilations.reduce((a,i)=>a+i,0)>0&&r.dilations.length!==n)throw new Error(`dilations should be ${n}D`);if(r.strides.reduce((a,i)=>a+i,0)>0&&r.strides.length!==n)throw new Error(`strides should be ${n}D`);if(r.pads.reduce((a,i)=>a+i,0)>0&&r.pads.length!==n*2)throw new Error(`pads should be ${n*2}D`);if(r.outputPadding.length!==n&&r.outputPadding.length!==0)throw new Error(`output_padding should be ${n}D`);if(r.kernelShape.reduce((a,i)=>a+i,0)>0&&r.kernelShape.length!==0&&r.kernelShape.length!==e[1].dims.length-2)throw new Error("invalid kernel shape");if(r.outputShape.length!==0&&r.outputShape.length!==e[0].dims.length-2)throw new Error("invalid output shape")},Bl=(e,r,t,s)=>{let o=e.kernelCustomData.wT??e.compute(rs(r[1],[2,3,0,1]),{inputs:[1],outputs:[t.wIsConst?-2:-1]})[0];t.wIsConst&&!e.kernelCustomData.wT&&(e.kernelCustomData.wT=o);let n=[r[0],o];r.length===3&&n.push(r[2]),e.compute(qm(n,t,s),{inputs:n})},ef=(e,r)=>{let t=r.format==="NHWC",s=[e.inputs[0].reshape(t?[e.inputs[0].dims[0],1,e.inputs[0].dims[1],e.inputs[0].dims[2]]:[e.inputs[0].dims[0],e.inputs[0].dims[1],1,e.inputs[0].dims[2]]),e.inputs[1].reshape([e.inputs[1].dims[0],e.inputs[1].dims[1],1,e.inputs[1].dims[2]])];e.inputs.length===3&&s.push(e.inputs[2]);let o=r.kernelShape;(o.length===0||o[0]===0)&&(o=[e.inputs[1].dims[2]]);let n=r.dilations;(n.length===0||n[0]===0)&&(n=[1]);let a=r.strides;(a.length===0||a[0]===0)&&(a=[1]);let i=r.pads;i.length===0&&(i=[0,0]),i=[0,i[0],0,i[1]],a=[1].concat(a),n=[1].concat(n),o=[1].concat(o);let l=r.outputPadding;l=[0].concat(l);let c=Rl({...r,pads:i,strides:a,dilations:n,kernelShape:o,outputPadding:l},s);Bl(e,s,c,p=>t?[p[0],p[2],p[3]]:[p[0],p[1],p[3]])},tf=(e,r)=>{if(Zm(e.inputs,r),e.inputs[0].dims.length===3)ef(e,r);else{let t=Rl(r,e.inputs);Bl(e,e.inputs,t)}}}),rf,sf,nf,lT=Ue(()=>{Mt(),Pt(),ur(),Ct(),rf=(e,r,t,s)=>{let o=Pe.size(r),n=r.length,a=Oe("input",e,n),i=at("output",e,n),l=t.dataType===6?t.getInt32Array()[0]:Number(t.getBigInt64Array()[0]),c=Pe.normalizeAxis(l,n),p=d=>{let u=` i32(${a.indicesGet("inputIndices","uniforms.axis")}) `,_=it("uniforms.input_shape","uniforms.axis",n),f=s.reverse?u+(s.exclusive?" + 1":""):"0",b=s.reverse?_:u+(s.exclusive?"":" + 1");return`
                ${d.registerUniform("outputSize","u32").registerUniform("axis","u32").declareVariables(a,i)}
                ${d.mainStart()}
                  ${d.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.outputSize")}
                  var inputIndices = ${i.offsetToIndices("global_idx")};
                  var sum = ${i.type.value}(0);
                  let first : i32 = ${f};
                  let last : i32 = ${b};
                  for (var i : i32 = first; i < last; i++) {
                    ${a.indicesSet("inputIndices","uniforms.axis","u32(i)")};
                    sum = sum + ${a.getByIndices("inputIndices")};
                  }
                  ${i.setByOffset("global_idx","sum")};
                }`};return{name:"CumSum",shaderCache:{hint:s.cacheKey,inputDependencies:["rank"]},getRunData:()=>({outputs:[{dims:r,dataType:e}],dispatchGroup:{x:Math.ceil(o/64)},programUniforms:[{type:12,data:o},{type:12,data:c},...lt(r,r)]}),getShaderSource:p}},sf=(e,r)=>{let t=e.inputs[0].dims,s=e.inputs[0].dataType,o=e.inputs[1];e.compute(rf(s,t,o,r),{inputs:[0]})},nf=e=>{let r=e.exclusive===1,t=e.reverse===1;return jt({exclusive:r,reverse:t})}}),of,af,lf,uf,cf,uT=Ue(()=>{Mt(),Pt(),ur(),Ct(),of=e=>{if(!e||e.length!==1)throw new Error("DepthToSpace requires 1 input.");if(e[0].dims.length!==4)throw new Error("DepthToSpace requires 4D input.")},af=(e,r,t,s)=>{let o=[];o.push(`fn perm(i: ${s.type.indices}) -> ${t.type.indices} {
    var a: ${t.type.indices};`);for(let n=0;n<r;++n)o.push(t.indicesSet("a",e[n],`i[${n}]`));return o.push("return a;}"),o.join(`
`)},lf=(e,r)=>{let t,s,o,n,a,i,l=r.format==="NHWC",c=r.blocksize,p=r.mode==="DCR";l?([t,s,o,n]=e.dims,a=p?[t,s,o,c,c,n/c**2]:[t,s,o,n/c**2,c,c],i=p?[0,1,3,2,4,5]:[0,1,4,2,5,3]):([t,s,o,n]=[e.dims[0],e.dims[2],e.dims[3],e.dims[1]],a=p?[t,c,c,n/c**2,s,o]:[t,n/c**2,c,c,s,o],i=p?[0,3,4,1,5,2]:[0,1,4,2,5,3]);let d=e.reshape(a),u=d.dims.length,_=e.dataType,f=Oe("a",_,u),b=at("output",_,u),A=g=>`
  ${g.registerUniform("output_size","u32").declareVariables(f,b)}

  ${af(i,u,f,b)}

  ${g.mainStart()}
    ${g.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_size")}

    let indices = ${b.offsetToIndices("global_idx")};
    let aIndices = perm(indices);

    ${b.setByOffset("global_idx",f.getByIndices("aIndices"))}
  }`;return{name:"DepthToSpace",shaderCache:{hint:`${e.dims};${r.blocksize};${r.mode}`,inputDependencies:["rank"]},getRunData:g=>{let y=l?[t,s*c,o*c,n/c**2]:[t,n/c**2,s*c,o*c],C=Pe.size(y),x=d.dims,M=Pe.sortBasedOnPerm(x,i);return{outputs:[{dims:y,dataType:g[0].dataType}],dispatchGroup:{x:Math.ceil(C/64)},programUniforms:[{type:12,data:C},...lt(x,M)]}},getShaderSource:A}},uf=(e,r)=>{of(e.inputs),e.compute(lf(e.inputs[0],r))},cf=e=>jt({blocksize:e.blocksize,mode:e.mode,format:e.format})}),Ma,vo,Nl,df,pf,hf,mf,jl,ff,_f,gf,cT=Ue(()=>{Mt(),Pt(),ur(),Ct(),Ma="[a-zA-Z]|\\.\\.\\.",vo="("+Ma+")+",Nl="^"+vo+"$",df="("+vo+",)*"+vo,pf="^"+df+"$",hf=class{constructor(e=-1){this.symbolToIndices=new Map,this.inputIndex=e}addSymbol(e,r){let t=this.symbolToIndices.get(e);t===void 0?t=[r]:t.push(r),this.symbolToIndices.set(e,t)}},mf=class{constructor(e,r){var o;this.equation=r,this.hasEllipsis=!1,this.symbolToInfo=new Map,this.lhs=new Array,this.outputDims=[];let[t,s]=r.includes("->")?r.split("->",2):[r,""];if(!t.match(RegExp(pf)))throw new Error("Invalid LHS term");if(t.split(",").forEach((n,a)=>{let i=e[a].dims.slice();if(!n.match(RegExp(Nl)))throw new Error("Invalid LHS term");let l=this.processTerm(n,!0,i,a);this.lhs.push(l)}),s==="")s+=[...this.symbolToInfo.entries()].filter(([n,a])=>a.count===1||n==="...").map(([n])=>n).join("");else if(!s.match(RegExp(vo)))throw new Error("Invalid RHS");(o=s.match(RegExp(Ma,"g")))==null||o.forEach(n=>{if(n==="...")this.outputDims=this.outputDims.concat(this.ellipsisDims);else{let a=this.symbolToInfo.get(n);if(a===void 0)throw new Error("Invalid RHS symbol");this.outputDims.push(a.dimValue)}}),this.rhs=this.processTerm(s,!1,this.outputDims)}addSymbol(e,r,t){let s=this.symbolToInfo.get(e);if(s!==void 0){if(s.dimValue!==r&&s.count!==1)throw new Error("Dimension mismatch");s.count++,s.inputIndices.push(t)}else s={count:1,dimValue:r,inputIndices:[t]};this.symbolToInfo.set(e,s)}processTerm(e,r,t,s=-1){let o=t.length,n=!1,a=[],i=0;if(!e.match(RegExp(Nl))&&!r&&e!=="")throw new Error("Invalid LHS term");let l=e.match(RegExp(Ma,"g")),c=new hf(s);return l==null||l.forEach((p,d)=>{if(p==="..."){if(n)throw new Error("Only one ellipsis is allowed per input term");n=!0;let u=o-l.length+1;if(u<0)throw new Error("Ellipsis out of bounds");if(a=t.slice(i,i+u),this.hasEllipsis){if(this.ellipsisDims.length!==a.length||this.ellipsisDims.toString()!==a.toString())throw new Error("Ellipsis dimensions mismatch")}else if(r)this.hasEllipsis=!0,this.ellipsisDims=a;else throw new Error("Ellipsis must be specified in the LHS");for(let _=0;_<a.length;_++){let f=String.fromCharCode(48+_);c.addSymbol(f,d+_),this.addSymbol(f,t[i++],s)}}else c.addSymbol(p,d+(this.hasEllipsis?this.ellipsisDims.length-1:0)),this.addSymbol(p,t[i++],s)}),c}},jl=e=>e+"_max",ff=(e,r,t,s)=>{let o=e.map(c=>c.length).map((c,p)=>Oe(`input${p}`,r,c)),n=Pe.size(s),a=at("output",r,s.length),i=[...t.symbolToInfo.keys()].filter(c=>!t.rhs.symbolToIndices.has(c)),l=c=>{let p=[],d="var prod = 1.0;",u="var sum = 0.0;",_="sum += prod;",f=[],b=[],A=[],g=[],y=t.symbolToInfo.size===t.rhs.symbolToIndices.size;t.symbolToInfo.forEach((x,M)=>{var T;if(t.rhs.symbolToIndices.has(M)){let v=(T=t.rhs.symbolToIndices.get(M))==null?void 0:T[0];v!==void 0&&t.lhs.forEach((P,F)=>{if(x.inputIndices.includes(F)){let D=P.symbolToIndices.get(M);if(D===void 0)throw new Error("Invalid symbol error");D.forEach(K=>{p.push(`${o[F].indicesSet(`input${F}Indices`,K,a.indicesGet("outputIndices",v))}`)})}})}else t.lhs.forEach((v,P)=>{if(x.inputIndices.includes(P)){let F=v.symbolToIndices.get(M);if(F===void 0)throw new Error("Invalid symbol error");F.forEach(D=>{f.push(`${o[P].indicesSet(`input${P}Indices`,D,`${M}`)}`)}),g.push(`prod *= ${o[P].getByIndices(`input${P}Indices`)};`)}}),b.push(`for(var ${M}: u32 = 0; ${M} < uniforms.${jl(M)}; ${M}++) {`),A.push("}")});let C=y?[...p,`let sum = ${o.map((x,M)=>x.getByIndices(`input${M}Indices`)).join(" * ")};`]:[...p,u,...b,...f,d,...g,_,...A];return`
            ${c.registerUniforms(i.map(x=>({name:`${jl(x)}`,type:"u32"}))).registerUniform("outputSize","u32").declareVariables(...o,a)}

            ${c.mainStart()}
            ${c.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.outputSize")}
            var outputIndices = ${a.offsetToIndices("global_idx")};
            ${o.map((x,M)=>`var input${M}Indices: ${o[M].type.indices};`).join(`
`)}
            ${C.join(`
`)};
            ${a.setByOffset("global_idx","sum")};
          }`};return{name:"Einsum",shaderCache:{hint:t.equation,inputDependencies:e.map(()=>"rank")},getRunData:()=>{let c=i.filter(d=>t.symbolToInfo.has(d)).map(d=>{var u;return{type:12,data:((u=t.symbolToInfo.get(d))==null?void 0:u.dimValue)||0}});c.push({type:12,data:n});let p=e.map((d,u)=>[...lt(d)]).reduce((d,u)=>d.concat(u),c);return p.push(...lt(s)),{outputs:[{dims:s,dataType:r}],dispatchGroup:{x:Math.ceil(n/64)},programUniforms:p}},getShaderSource:l}},_f=(e,r)=>{let t=new mf(e.inputs,r.equation),s=t.outputDims,o=e.inputs.map((n,a)=>n.dims);e.compute(ff(o,e.inputs[0].dataType,t,s))},gf=e=>{let r=e.equation.replace(/\s+/g,"");return jt({equation:r})}}),wf,Vl,Mf,bf,yf,dT=Ue(()=>{Mt(),Pt(),Ct(),wf=e=>{if(!e||e.length!==2)throw new Error("Expand requires 2 input.");let r=e[0].dims,t=Array.from(e[1].getBigInt64Array(),Number),s=t.length<r.length?0:t.length-r.length,o=r.length<t.length?0:r.length-t.length;for(;s<t.length&&o<r.length;++s,++o)if(t[s]!==r[o]&&t[s]!==1&&r[o]!==1)throw new Error("Expand requires shape to be broadcastable to input")},Vl=(e,r)=>{let t=e.length-r.length,s=[];for(let o=0;o<t;++o)s.push(e[o]);for(let o=0;o<r.length;++o)s.push(r[o]===1?e[o+t]:r[o]);return s},Mf=(e,r)=>e.length>r.length?Vl(e,r):Vl(r,e),bf=e=>{let r=e[0].dims,t=Array.from(e[1].getBigInt64Array(),Number),s=Mf(r,t),o=e[0].dataType,n=o===9||Pe.size(r)===1,a=o===9||r.length>0&&r[r.length-1]%4===0?4:1,i=n||s.length>0&&s[s.length-1]%4===0?4:1,l=Math.ceil(Pe.size(s)/i),c=d=>{let u=Oe("input",o,r.length,a),_=at("output",o,s.length,i),f;if(o===9){let b=(A,g,y="")=>`
          let outputIndices${g} = ${_.offsetToIndices(`outputOffset + ${g}u`)};
          let offset${g} = ${u.broadcastedIndicesToOffset(`outputIndices${g}`,_)};
          let index${g} = offset${g} / 4u;
          let component${g} = offset${g} % 4u;
          ${A}[${g}] = ${y}(${u.getByOffset(`index${g}`)}[component${g}]);
        `;f=`
        let outputOffset = global_idx * ${i};
        var data = vec4<u32>(0);
        ${b("data",0,"u32")}
        ${b("data",1,"u32")}
        ${b("data",2,"u32")}
        ${b("data",3,"u32")}
        ${_.setByOffset("global_idx","data")}
      }`}else f=`
        let outputIndices = ${_.offsetToIndices(`global_idx * ${i}`)};
        let inputOffset = ${u.broadcastedIndicesToOffset("outputIndices",_)};
        let data = ${_.type.value}(${u.getByOffset(`inputOffset / ${a}`)});
        ${_.setByOffset("global_idx","data")}
      }`;return`
    ${d.registerUniform("vec_size","u32").declareVariables(u,_)}
    ${d.mainStart()}
    ${d.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.vec_size")}
    ${f}`},p=[{type:12,data:l},...lt(r,s)];return{name:"Expand",shaderCache:{hint:`${s.length};${a}${i}`,inputDependencies:["rank"]},getShaderSource:c,getRunData:()=>({outputs:[{dims:s,dataType:e[0].dataType}],dispatchGroup:{x:Math.ceil(l/64)},programUniforms:p})}},yf=e=>{wf(e.inputs),e.compute(bf(e.inputs),{inputs:[0]})}}),vf,xf,pT=Ue(()=>{Mt(),Pt(),Ct(),Tl(),vf=e=>{let r=e[0].dataType,t=Pe.size(e[0].dims),s=Pe.size(e[1].dims),o=s%4===0,n=a=>{let i=Oe("x",r,[1],4),l=Oe("bias",r,[1],4),c=at("y",r,[1],4),p=[{name:"output_vec_size",type:"u32"},{name:"bias_size",type:"u32"}],d=_=>`
      let bias${_}_offset: u32 = (global_idx * 4 + ${_}) % uniforms.bias_size;
      let bias${_} = ${l.getByOffset(`bias${_}_offset / 4`)}[bias${_}_offset % 4];`,u=o?`
      let bias = ${l.getByOffset("global_idx % (uniforms.bias_size / 4)")};`:`${d(0)}${d(1)}${d(2)}${d(3)}
      let bias = ${i.type.value}(bias0, bias1, bias2, bias3);`;return`${a.registerUniforms(p).declareVariables(i,l,c)}

    ${vl(jr(r))}

    ${a.mainStart(Hn)}
      ${a.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_vec_size")}

      let x = ${i.getByOffset("global_idx")};
      ${u}
      let x_in = x + bias;
      ${c.setByOffset("global_idx",xl("x_in"))}
    }`};return{name:"FastGeluWithBias",shaderCache:{hint:`${o}`,inputDependencies:["type","type"]},getShaderSource:n,getRunData:a=>({outputs:[{dims:a[0].dims,dataType:a[0].dataType}],programUniforms:[{type:12,data:Math.ceil(t/4)},{type:12,data:s}],dispatchGroup:{x:Math.ceil(t/Hn/4)}})}},xf=e=>{e.inputs.length<2||Pe.size(e.inputs[1].dims)===0?rm(e):e.compute(vf(e.inputs))}}),Tf,Ef,Pf,Cf,hT=Ue(()=>{Mt(),Pt(),ur(),Ct(),Tf=e=>{if(!e||e.length!==2)throw new Error("Gather requires 2 inputs.")},Ef=(e,r)=>{let t=e[0].dims,s=e[1].dims,o=t.length,n=Pe.normalizeAxis(r.axis,o),a=t.slice(0);a.splice(n,1,...s);let i=t[n],l=e[0].dataType===9?4:1,c=Math.ceil(Pe.size(a)/l),p=[{type:12,data:c},{type:6,data:i},{type:12,data:n},...lt(e[0].dims,e[1].dims,a)],d=u=>{let _=Oe("data",e[0].dataType,e[0].dims.length,l),f=Oe("inputIndices",e[1].dataType,e[1].dims.length),b=at("output",e[0].dataType,a.length,l),A=y=>{let C=s.length,x=`var indicesIndices${y}  = ${f.type.indices}(0);`;for(let M=0;M<C;M++)x+=`${C>1?`indicesIndices${y}[${M}]`:`indicesIndices${y}`} = ${a.length>1?`outputIndices${y}[uniforms.axis + ${M}]`:`outputIndices${y}`};`;x+=`
          var idx${y} = ${f.getByIndices(`indicesIndices${y}`)};
          if (idx${y} < 0) {
            idx${y} = idx${y} + uniforms.axisDimLimit;
          }
          var dataIndices${y} : ${_.type.indices};
        `;for(let M=0,T=0;M<o;M++)M===n?(x+=`${o>1?`dataIndices${y}[${M}]`:`dataIndices${y}`} = u32(idx${y});`,T+=C):(x+=`${o>1?`dataIndices${y}[${M}]`:`dataIndices${y}`} = ${a.length>1?`outputIndices${y}[${T}]`:`outputIndices${y}`};`,T++);return x},g;if(e[0].dataType===9){let y=(C,x,M="")=>`
          let outputIndices${x} = ${b.offsetToIndices(`outputOffset + ${x}u`)};
          ${A(x)};
          let offset${x} = ${_.indicesToOffset(`dataIndices${x}`)};
          let index${x} = offset${x} / 4u;
          let component${x} = offset${x} % 4u;
          ${C}[${x}] = ${M}(${_.getByOffset(`index${x}`)}[component${x}]);
        `;g=`
        let outputOffset = global_idx * ${l};
        var value = vec4<u32>(0);
        ${y("value",0,"u32")}
        ${y("value",1,"u32")}
        ${y("value",2,"u32")}
        ${y("value",3,"u32")}
        ${b.setByOffset("global_idx","value")}
      `}else g=`
      let outputIndices = ${b.offsetToIndices("global_idx")};
      ${A("")};
      let value = ${_.getByIndices("dataIndices")};
      ${b.setByOffset("global_idx","value")};
      `;return`
      ${u.registerUniform("outputSize","u32").registerUniform("axisDimLimit","i32").registerUniform("axis","u32").declareVariables(_,f,b)}
      ${u.mainStart()}
        ${u.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.outputSize")}
        ${g}
      }`};return{name:"Gather",shaderCache:{hint:r.cacheKey,inputDependencies:["rank","rank"]},getRunData:()=>({outputs:[{dims:a,dataType:e[0].dataType}],dispatchGroup:{x:Math.ceil(c/64)},programUniforms:p}),getShaderSource:d}},Pf=e=>jt({axis:e.axis}),Cf=(e,r)=>{let t=e.inputs;Tf(t),e.compute(Ef(e.inputs,r))}}),Sf,$f,Af,mT=Ue(()=>{Mt(),Pt(),Ct(),Sf=(e,r,t,s,o,n,a,i,l)=>{let c=[{type:12,data:n},{type:12,data:s},{type:12,data:o},{type:12,data:t},{type:12,data:a},{type:12,data:i},{type:12,data:l}],p=[n];c.push(...lt(r.dims,p));let d=u=>{let _=Oe("indices_data",r.dataType,r.dims.length),f=at("input_slice_offsets_data",12,1,1),b=[_,f],A=[{name:"output_size",type:"u32"},{name:"batch_dims",type:"u32"},{name:"input_dims",type:"u32",length:o.length},{name:"sizes_from_slice_dims_data",type:"u32",length:t.length},{name:"num_slices_per_batch",type:"u32"},{name:"input_batch_stride",type:"u32"},{name:"num_slice_dims",type:"u32"}];return`
  ${u.registerUniforms(A).declareVariables(...b)}
  ${u.mainStart()}
    ${u.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_size")}
    let batch_idx = global_idx / uniforms.num_slices_per_batch;
    let base_offset = batch_idx * uniforms.input_batch_stride;

    let slice_indices_base_offset = global_idx * uniforms.num_slice_dims;
    var relative_slice_offset = 0;
    for (var dim_idx = 0u; dim_idx < uniforms.num_slice_dims; dim_idx ++) {
      var index = i32(indices_data[dim_idx + slice_indices_base_offset].x);
      let input_dim_idx = uniforms.batch_dims + dim_idx;
      if (index < 0) {
        ${o.length===1?"index += i32(uniforms.input_dims);":"index += i32(uniforms.input_dims[input_dim_idx]);"}
      }
      ${t.length===1?"relative_slice_offset += index * i32(uniforms.sizes_from_slice_dims_data);":"relative_slice_offset += index * i32(uniforms.sizes_from_slice_dims_data[dim_idx]);"}
    }

    input_slice_offsets_data[global_idx] =  base_offset + u32(relative_slice_offset);
  }`};return e.compute({name:"computeSliceOffsets",shaderCache:{hint:`${o.length}_${t.length}`,inputDependencies:["rank"]},getRunData:()=>({outputs:[{dims:p,dataType:e.inputs[1].dataType}],dispatchGroup:{x:Math.ceil(n/64)},programUniforms:c}),getShaderSource:d},{inputs:[r],outputs:[-1]})[0]},$f=(e,r)=>{let t=e.inputs,s=t[0].dims,o=t[0].dataType,n=t[1].dims,a=n[n.length-1],i=Pe.sizeToDimension(n,n.length-1),l=Pe.sizeFromDimension(s,r.batchDims+a),c=Pe.sizeToDimension(s,r.batchDims),p=Pe.sizeFromDimension(s,r.batchDims),d=i/c,u=new Array(a),_=l;for(let x=0;x<a;++x)u[a-1-x]=_,_*=s[r.batchDims+a-1-x];let f=Sf(e,t[1],u,r.batchDims,s,i,d,p,a),b=r.batchDims+a;if(b>s.length)throw new Error("last dimension of indices must not be larger than rank of input tensor");let A=n.slice(0,-1).concat(s.slice(b)),g=Pe.size(A),y=[{type:12,data:g},{type:12,data:l},...lt(t[0].dims,f.dims,A)],C=x=>{let M=Oe("data",t[0].dataType,t[0].dims.length),T=Oe("slice_offsets",12,f.dims.length),v=at("output",t[0].dataType,A.length);return`
          ${x.registerUniform("output_size","u32").registerUniform("slice_size","u32").declareVariables(M,T,v)}
            ${x.mainStart()}
            ${x.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_size")}
          let slice_offset = slice_offsets[global_idx / uniforms.slice_size];
          output[global_idx] = data[u32(slice_offset) + global_idx % uniforms.slice_size];
        }`};e.compute({name:"GatherND",shaderCache:{hint:r.cacheKey,inputDependencies:["rank","rank"]},getRunData:()=>({outputs:[{dims:A,dataType:o}],dispatchGroup:{x:Math.ceil(g/64)},programUniforms:y}),getShaderSource:C},{inputs:[t[0],f]})},Af=e=>({batchDims:e.batch_dims,cacheKey:""})}),kf,If,Ff,Of,fT=Ue(()=>{Mt(),Pt(),ur(),Ct(),kf=(e,r)=>{if(e.length<3||e.length>4)throw new Error("GatherBlockQuantized requires 3 or 4 inputs.");let t=Pe.normalizeAxis(r.quantizeAxis,e[0].dims.length),s=r.blockSize,o=e[0],n=e[2],a=e.length===4?e[3]:void 0;if(n.dims.length!==o.dims.length||!o.dims.map((i,l)=>l===t?Math.ceil(i/s)===n.dims[l]:i===n.dims[l]).reduce((i,l)=>i&&l,!0))throw new Error("Scales must have the same rank as the input tensor and the dims should match except on gatherAxis.");if(a){if(a.dataType!==o.dataType)throw new Error("Zero point must have the same data type as the input tensor.");if(a.dims.length!==n.dims.length||!a.dims.map((i,l)=>i===n.dims[l]).reduce((i,l)=>i&&l,!0))throw new Error("Zero point must have the same rank as the input tensor and the dims should match except on quantizeAxis.")}},If=(e,r)=>{let t=e[0].dims,s=e[1].dims,o=t.length,n=Pe.normalizeAxis(r.gatherAxis,o),a=Pe.normalizeAxis(r.quantizeAxis,o),i=t.slice(0);i.splice(n,1,...s);let l=Pe.size(i),c=e[2].dataType,p=e[0].dataType===22,d=[{type:12,data:l},{type:12,data:a},{type:12,data:n},{type:12,data:r.blockSize},...lt(...e.map((_,f)=>_.dims),i)],u=_=>{let f=Oe("data",e[0].dataType,e[0].dims.length),b=Oe("inputIndices",e[1].dataType,e[1].dims.length),A=Oe("scales",e[2].dataType,e[2].dims.length),g=e.length>3?Oe("zeroPoint",e[3].dataType,e[3].dims.length):void 0,y=at("output",c,i.length),C=[f,b,A];g&&C.push(g);let x=[{name:"output_size",type:"u32"},{name:"quantize_axis",type:"u32"},{name:"gather_axis",type:"u32"},{name:"block_size",type:"u32"}];return`
        ${_.registerUniforms(x).declareVariables(...C,y)}
        ${_.mainStart()}
        let output_indices = ${y.offsetToIndices("global_idx")};
        var indices_indices = ${b.type.indices}(0);
        ${s.length>1?`
          for (var i: u32 = 0; i < ${s.length}; i++) {
            let index = ${y.indicesGet("output_indices","uniforms.gather_axis + i")};
            ${b.indicesSet("indices_indices","i","index")};
          }`:`indices_indices = ${y.indicesGet("output_indices","uniforms.gather_axis")};`};
        var data_indices = ${f.type.indices}(0);
        for (var i: u32 = 0; i < uniforms.gather_axis; i++) {
          let index = ${y.indicesGet("output_indices","i")};
          ${f.indicesSet("data_indices","i","index")};
        }
        var index_from_indices = ${b.getByIndices("indices_indices")};
        if (index_from_indices < 0) {
          index_from_indices += ${t[n]};
        }
        ${f.indicesSet("data_indices","uniforms.gather_axis","u32(index_from_indices)")};
        for (var i = uniforms.gather_axis + 1; i < ${i.length}; i++) {
          let index = ${y.indicesGet("output_indices",`i + ${s.length} - 1`)};
          ${f.indicesSet("data_indices","i","index")};
        }
        let data_offset = ${f.indicesToOffset("data_indices")};
        let data_index = data_offset % 8;
        // Convert 4-bit packed data to 8-bit packed data.
        let packed_4bit_quantized_data = ${f.getByOffset("data_offset / 8")};
        let packed_8bit_quantized_data = (packed_4bit_quantized_data >> (4 * (data_index % 2))) & 0x0f0f0f0f;
        let quantized_data_vec = ${p?"unpack4xI8":"unpack4xU8"}(u32(packed_8bit_quantized_data));
        let quantized_data = quantized_data_vec[data_index / 2];
        var scale_indices = data_indices;
        let quantize_axis_index = ${A.indicesGet("data_indices","uniforms.quantize_axis")} / uniforms.block_size;
        ${A.indicesSet("scale_indices","uniforms.quantize_axis","quantize_axis_index")};
        var scale = ${A.getByIndices("scale_indices")};
        ${g?`
              let zero_point_indices = scale_indices;
              let zero_point_offset = ${g.indicesToOffset("zero_point_indices")};
              let zero_point_index = zero_point_offset % 8;
              let packed_4bit_zero_points = ${g.getByOffset("zero_point_offset / 8")};
              let packed_8bit_zero_points = (packed_4bit_zero_points >> (4 * (zero_point_index % 2))) & 0x0f0f0f0f;
              let zero_point_vec = ${p?"unpack4xI8":"unpack4xU8"}(u32(packed_8bit_zero_points));
              let zero_point = zero_point_vec[zero_point_index / 2];`:"var zero_point = 0"};
        let dequantized_data = ${jr(c)}(quantized_data - zero_point) * scale;
        ${y.setByOffset("global_idx","dequantized_data")};
    }`};return{name:"GatherBlockQuantized",shaderCache:{hint:`${r.cacheKey};${e.filter((_,f)=>f!==1).map(_=>_.dims.join("_")).join(";")}`,inputDependencies:Array.from({length:e.length},(_,f)=>"rank")},getRunData:()=>({outputs:[{dims:i,dataType:c}],dispatchGroup:{x:Math.ceil(l/64)},programUniforms:d}),getShaderSource:u}},Ff=(e,r)=>{let t=e.inputs;kf(t,r),e.compute(If(e.inputs,r))},Of=e=>jt({blockSize:e.blockSize,gatherAxis:e.gatherAxis,quantizeAxis:e.quantizeAxis})}),Df,Lf,zf,Rf,_T=Ue(()=>{Mt(),Pt(),ur(),Ct(),Df=e=>{if(!e||e.length!==2)throw new Error("GatherElements requires 2 inputs.");if(e[0].dims.length<1)throw new Error("GatherElements requires that the data input be rank >= 1.");if(e[0].dims.length!==e[1].dims.length)throw new Error(`GatherElements requires that the data input and
                     indices input tensors be of same rank.`)},Lf=(e,r)=>{let t=e[0].dims,s=e[0].dataType,o=t.length,n=e[1].dims,a=e[1].dataType,i=Pe.normalizeAxis(r.axis,o),l=t[i],c=n.slice(0),p=Pe.size(c),d=Oe("input",s,o),u=Oe("indicesInput",a,n.length),_=at("output",s,c.length),f=[{type:12,data:p},{type:6,data:l},{type:12,data:i}];return f.push(...lt(t,n,c)),{name:"GatherElements",shaderCache:{inputDependencies:["rank","rank"]},getRunData:()=>({outputs:[{dims:c,dataType:e[0].dataType}],dispatchGroup:{x:Math.ceil(p/64)},programUniforms:f}),getShaderSource:b=>`
      ${b.registerUniform("outputSize","u32").registerUniform("axisDimLimit","i32").registerUniform("axis","u32").declareVariables(d,u,_)}
      ${b.mainStart()}
      ${b.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.outputSize")}

      let outputIndices = ${_.offsetToIndices("global_idx")};

      var idx = ${u.getByOffset("global_idx")};
      if (idx < 0) {
        idx = idx + uniforms.axisDimLimit;
      }
      var inputIndices = ${d.type.indices}(outputIndices);
      ${d.indicesSet("inputIndices","uniforms.axis","u32(idx)")};
      let value = ${d.getByIndices("inputIndices")};

      ${_.setByOffset("global_idx","value")};
  }`}},zf=e=>jt({axis:e.axis}),Rf=(e,r)=>{let t=e.inputs;Df(t),e.compute(Lf(e.inputs,r))}}),Bf,Nf,jf,Vf,gT=Ue(()=>{Mt(),Pt(),Ct(),Bf=e=>{if(!e)throw new Error("Input is missing");if(e.length<2||e.length>3)throw new Error("Invaid input number.");if(e.length===3&&e[2].dims.length>2)throw new Error("Invalid input shape of C");if(e[0].dataType!==e[1].dataType||e.length===3&&e[0].dataType!==e[2].dataType)throw new Error("Input types are mismatched")},Nf=(e,r)=>{let t=e[0].dims.slice(),s=e[1].dims.slice(),[o,n,a]=Xd.getShapeOfGemmResult(t,r.transA,s,r.transB,e.length===3?e[2].dims:void 0),i=[o,n];if(!i)throw new Error("Can't use gemm on the given tensors");let l=16,c=Math.ceil(n/l),p=Math.ceil(o/l),d=!0,u=Pe.size(i),_=[{type:12,data:d?c:u},{type:12,data:o},{type:12,data:n},{type:12,data:a},{type:1,data:r.alpha},{type:1,data:r.beta}],f=["type","type"];e.length===3&&(_.push(...lt(e[2].dims)),f.push("rank")),_.push(...lt(i));let b=g=>{let y="";r.transA&&r.transB?y="value += a[k * uniforms.M + m] * b[n * uniforms.K + k];":r.transA&&!r.transB?y="value += a[k * uniforms.M + m] * b[k * uniforms.N + n];":!r.transA&&r.transB?y="value += a[m * uniforms.K + k] * b[n * uniforms.K + k];":!r.transA&&!r.transB&&(y="value += a[m * uniforms.K + k] * b[k * uniforms.N + n];");let C=r.alpha===1?"":"value *= uniforms.alpha;",x=Oe("a",e[0].dataType,e[0].dims),M=Oe("b",e[1].dataType,e[1].dims),T=x.type.value,v=null,P=[x,M];e.length===3&&(v=Oe("c",e[2].dataType,e[2].dims.length),P.push(v));let F=at("output",e[0].dataType,i.length);P.push(F);let D=[{name:"output_size",type:"u32"},{name:"M",type:"u32"},{name:"N",type:"u32"},{name:"K",type:"u32"},{name:"alpha",type:"f32"},{name:"beta",type:"f32"}];return`
  ${g.registerUniforms(D).declareVariables(...P)}

  ${g.mainStart()}
    ${g.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_size")}

    let m = global_idx / uniforms.N;
    let n = global_idx % uniforms.N;

    var value = ${T}(0);
    for (var k: u32 = 0u; k < uniforms.K; k++) {
      ${y}
    }

    ${C}
    ${v!=null?`let cOffset = ${v.broadcastedIndicesToOffset("vec2(m, n)",F)}; value += ${T}(uniforms.beta) * ${v.getByOffset("cOffset")};`:""}
    output[global_idx] = value;
  }`},A=g=>{let y=Oe("a",e[0].dataType,e[0].dims),C=Oe("b",e[1].dataType,e[1].dims),x=null,M=[y,C];e.length===3&&(x=Oe("c",e[2].dataType,e[2].dims.length),M.push(x));let T=at("output",e[0].dataType,i.length);M.push(T);let v=[{name:"num_tile_n",type:"u32"},{name:"M",type:"u32"},{name:"N",type:"u32"},{name:"K",type:"u32"},{name:"alpha",type:"f32"},{name:"beta",type:"f32"}],P="",F="";r.transA&&r.transB?(F=`
      var col = tile_row_start + local_id.x;
      var row = k_start + local_id.y;
      if (col < uniforms.M && row < uniforms.K) {
        tile_a[local_id.y][local_id.x] = a[row * uniforms.M + col];
      } else {
        tile_a[local_id.y][local_id.x] = ${y.type.value}(0);
      }

      col = k_start + local_id.x;
      row = tile_col_start + local_id.y;
      if (col < uniforms.K && row < uniforms.N) {
        tile_b[local_id.y][local_id.x] = b[row * uniforms.K + col];
      } else {
        tile_b[local_id.y][local_id.x] = ${C.type.value}(0);
      }
      `,P="value += tile_a[k][local_id.y] * tile_b[local_id.x][k];"):r.transA&&!r.transB?(F=`
      var col = tile_row_start + local_id.x;
      var row = k_start + local_id.y;
      if (col < uniforms.M && row < uniforms.K) {
        tile_a[local_id.y][local_id.x] = a[row * uniforms.M + col];
      } else {
        tile_a[local_id.y][local_id.x] = ${y.type.value}(0);
      }

      col = tile_col_start + local_id.x;
      row = k_start + local_id.y;
      if (col < uniforms.N && row < uniforms.K) {
        tile_b[local_id.y][local_id.x] = b[row * uniforms.N + col];
      } else {
        tile_b[local_id.y][local_id.x] = ${C.type.value}(0);
      }
      `,P="value += tile_a[k][local_id.y] * tile_b[k][local_id.x];"):!r.transA&&r.transB?(F=`
      var col = k_start + local_id.x;
      var row = tile_row_start + local_id.y;
      if (col < uniforms.K && row < uniforms.M) {
        tile_a[local_id.y][local_id.x] = a[row * uniforms.K + col];
      } else {
        tile_a[local_id.y][local_id.x] = ${y.type.value}(0);
      }

      col = k_start + local_id.x;
      row = tile_col_start + local_id.y;
      if (col < uniforms.K && row < uniforms.N) {
        tile_b[local_id.y][local_id.x] = b[row * uniforms.K + col];
      } else {
        tile_b[local_id.y][local_id.x] = ${C.type.value}(0);
      }
      `,P="value += tile_a[local_id.y][k] * tile_b[local_id.x][k];"):!r.transA&&!r.transB&&(F=`
      var col = k_start + local_id.x;
      var row = tile_row_start + local_id.y;
      if (col < uniforms.K && row < uniforms.M) {
        tile_a[local_id.y][local_id.x] = a[row * uniforms.K + col];
      } else {
        tile_a[local_id.y][local_id.x] = ${y.type.value}(0);
      }

      col = tile_col_start + local_id.x;
      row = k_start + local_id.y;
      if (col < uniforms.N && row < uniforms.K) {
        tile_b[local_id.y][local_id.x] = b[row * uniforms.N + col];
      } else {
        tile_b[local_id.y][local_id.x] = ${C.type.value}(0);
      }
      `,P="value += tile_a[local_id.y][k] * tile_b[k][local_id.x];");let D=r.alpha===1?"":"value *= uniforms.alpha;";return`
  ${g.registerUniforms(v).declareVariables(...M)}
  var<workgroup> tile_a: array<array<${y.type.storage}, ${l}>, ${l}>;
  var<workgroup> tile_b: array<array<${C.type.storage}, ${l}>, ${l}>;
  ${g.mainStart([l,l,1])}
    let tile_col_start = (workgroup_index % uniforms.num_tile_n) * ${l};
    let tile_row_start = (workgroup_index / uniforms.num_tile_n) * ${l};
    let num_tiles = (uniforms.K - 1) / ${l} + 1;
    var k_start = 0u;
    var value = ${T.type.value}(0);
    for (var t: u32 = 0u; t < num_tiles; t++) {
      ${F}
      k_start = k_start + ${l};
      workgroupBarrier();

      for (var k: u32 = 0u; k < ${l}; k++) {
        ${P}
      }
      workgroupBarrier();
    }

    ${D}
    let m = tile_row_start + local_id.y;
    let n = tile_col_start + local_id.x;
    ${x!=null?`let cOffset = ${x.broadcastedIndicesToOffset("vec2(m, n)",T)}; value += ${T.type.value}(uniforms.beta) * ${x.getByOffset("cOffset")};`:""}
    if (m < uniforms.M && n < uniforms.N) {
      output[m * uniforms.N + n] = value;
    }
  }`};return d?{name:"GemmShared",shaderCache:{hint:`${r.cacheKey}`,inputDependencies:f},getRunData:()=>({outputs:[{dims:i,dataType:e[0].dataType}],dispatchGroup:{x:c*p},programUniforms:_}),getShaderSource:A}:{name:"Gemm",shaderCache:{hint:`${r.cacheKey}`,inputDependencies:f},getRunData:()=>({outputs:[{dims:i,dataType:e[0].dataType}],dispatchGroup:{x:Math.ceil(u/64)},programUniforms:_}),getShaderSource:b}},jf=e=>{let r=e.transA,t=e.transB,s=e.alpha,o=e.beta;return{transA:r,transB:t,alpha:s,beta:o,cacheKey:`${e.transA};${e.transB};${e.alpha===1}`}},Vf=(e,r)=>{Bf(e.inputs),e.compute(Nf(e.inputs,r))}}),Fs,Vs,bn,yn,Uf,Wf,Gf,Hf,Kf,qf,Qf,Xf,Jf,Yf,wT=Ue(()=>{Mt(),Pt(),ur(),Ct(),[Fs,Vs,bn,yn]=[0,1,2,3],Uf=e=>{if(e[0].dims.length!==4)throw new Error("only 4-D tensor is supported.");if(e[0].dims.length!==e[1].dims.length)throw new Error("input dimensions must be equal to grid dimensions");if(e[0].dims.length-2!==e[1].dims[e[1].dims.length-1])throw new Error(`last dimension of grid must be equal to ${e[0].dims.length-2}`);if(e[0].dims[0]!==e[1].dims[0])throw new Error("grid batch size must match input batch size")},Wf=`
  fn gs_get_cubic_coeffs(x: f32) -> vec4<f32> {
    let cubic_alpha = -0.75f;
    let x_abs = abs(x);
    var coeffs: vec4<f32>;
    coeffs[0] = (((cubic_alpha * (x_abs + 1) - 5 * cubic_alpha) * (x_abs + 1) + 8 * cubic_alpha) * (x_abs + 1) - 4 * cubic_alpha);
    coeffs[1] = (((cubic_alpha + 2) * x_abs - (cubic_alpha + 3)) * x_abs * x_abs + 1);
    coeffs[2] = (((cubic_alpha + 2) * (1 - x_abs) - (cubic_alpha + 3)) * (1 - x_abs) * (1 - x_abs) + 1);
    coeffs[3] = (((cubic_alpha * (2 - x_abs) - 5 * cubic_alpha) * (2 - x_abs) + 8 * cubic_alpha) * (2 - x_abs) - 4 * cubic_alpha);
    return coeffs;
  }
`,Gf=e=>`
  fn gs_bicubic_interpolate(p: mat4x4<${e}>, x: f32, y: f32) -> ${e} {
    var v: vec4<f32>;
    var coeffs = gs_get_cubic_coeffs(x);
    for (var i = 0; i < 4; i++) {
      v[i] = coeffs[0] * p[i][0] + coeffs[1] * p[i][1] + coeffs[2] * p[i][2] + coeffs[3] * p[i][3];
    }
    coeffs = gs_get_cubic_coeffs(y);
    let pixel = ${e}(coeffs[0] * v[0] + coeffs[1] * v[1] + coeffs[2] * v[2] + coeffs[3] * v[3]);
    return pixel;
  }
`,Hf=e=>`
  fn gs_denormalize(n: f32, length: i32) -> f32 {
    ${e.alignCorners===0?`
    // alignCorners: false => [-1, 1] to [-0.5, length - 0.5]
    return ((n + 1.0) * f32(length) - 1.0) / 2.0;
    `:`
    // alignCorners: true => [-1, 1] to [0, length - 1]
    return (n + 1.0) / 2.0 * (f32(length - 1));
    `}
  }
`,Kf=e=>`
  ${e.paddingMode==="reflection"?`
      fn gs_reflect(x: i32, x_min: f32, x_max: f32) -> u32 {
        var dx = 0.0;
        var fx = f32(x);
        let range = x_max - x_min;
        if (fx < x_min) {
          dx = x_min - fx;
          let n = u32(dx / range);
          let r = dx - f32(n) * range;
          if (n % 2 == 0) {
            fx = x_min + r;
          } else {
            fx = x_max - r;
          }
        } else if (fx > x_max) {
          dx = fx - x_max;
          let n = u32(dx / range);
          let r = dx - f32(n) * range;
          if (n % 2 == 0) {
            fx = x_max - r;
          } else {
            fx = x_min + r;
          }
        }
        return u32(fx);
      }`:""}
`,qf=(e,r,t)=>`
  fn pixel_at_grid(r: i32, c: i32, H: i32, W: i32, batch: u32, channel: u32, border: vec4<f32>) -> ${r} {
     var pixel = ${r}(0);
     var indices = vec4<u32>(0);
     indices[${Fs}] = batch;
     indices[${Vs}] = channel;`+(()=>{switch(t.paddingMode){case"zeros":return`
          if (r >= 0 && r < H && c >=0 && c < W) {
            indices[${bn}] = u32(r);
            indices[${yn}] = u32(c);
          } else {
            return ${r}(0);
          }
        `;case"border":return`
          indices[${bn}] = u32(clamp(r, 0, H - 1));
          indices[${yn}] = u32(clamp(c, 0, W - 1));
        `;case"reflection":return`
          indices[${bn}] = gs_reflect(r, border[1], border[3]);
          indices[${yn}] = gs_reflect(c, border[0], border[2]);
        `;default:throw new Error(`padding mode ${t.paddingMode} is not supported`)}})()+`
    return ${e.getByIndices("indices")};
  }
`,Qf=(e,r,t)=>(()=>{switch(t.mode){case"nearest":return`
          let result = pixel_at_grid(i32(round(y)), i32(round(x)), H_in, W_in, indices[${Fs}], indices[${Vs}], border);
        `;case"bilinear":return`
          let x1 = i32(floor(x));
          let y1 = i32(floor(y));
          let x2 = x1 + 1;
          let y2 = y1 + 1;

          let p11 = pixel_at_grid(y1, x1, H_in, W_in, indices[${Fs}], indices[${Vs}], border);
          let p12 = pixel_at_grid(y1, x2, H_in, W_in, indices[${Fs}], indices[${Vs}], border);
          let p21 = pixel_at_grid(y2, x1, H_in, W_in, indices[${Fs}], indices[${Vs}], border);
          let p22 = pixel_at_grid(y2, x2, H_in, W_in, indices[${Fs}], indices[${Vs}], border);

          let dx2 = ${r}(f32(x2) - x);
          let dx1 = ${r}(x - f32(x1));
          let dy2 = ${r}(f32(y2) - y);
          let dy1 = ${r}(y - f32(y1));
          let result = dy2 * (dx2 * p11 + dx1 * p12) + dy1 * (dx2 * p21 + dx1 * p22);
        `;case"bicubic":return`
          let x0 = i32(floor(x)) - 1;
          let y0 = i32(floor(y)) - 1;
          var p: mat4x4<${r}>;
          for (var h = 0; h < 4; h++) {
            for (var w = 0; w < 4; w++) {
              p[h][w] = pixel_at_grid(h + y0, w + x0, H_in, W_in, indices[${Fs}], indices[${Vs}], border);
            }
          }

          let dx = x - f32(x0 + 1);
          let dy = y - f32(y0 + 1);
          let result = gs_bicubic_interpolate(p, dx, dy);
        `;default:throw new Error(`mode ${t.mode} is not supported`)}})()+`${e.setByOffset("global_idx","result")}`,Xf=(e,r)=>{let t=Oe("x",e[0].dataType,e[0].dims.length),s=[e[1].dims[0],e[1].dims[1],e[1].dims[2]],o=Oe("grid",e[1].dataType,s.length,2),n=[e[0].dims[0],e[0].dims[1],e[1].dims[1],e[1].dims[2]];r.format==="NHWC"&&(n=[e[0].dims[0],e[1].dims[1],e[1].dims[2],e[0].dims[3]],[Fs,Vs,bn,yn]=[0,3,1,2]);let a=at("output",e[0].dataType,n.length),i=t.type.value,l=Pe.size(n),c=[{type:12,data:l},...lt(e[0].dims,s,n)],p=d=>`
  ${d.registerUniform("output_size","u32").declareVariables(t,o,a)}
  ${Wf}
  ${Gf(i)}
  ${Hf(r)}
  ${Kf(r)}
  ${qf(t,i,r)}

  ${d.mainStart()}
    ${d.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_size")}
      let H_in = i32(uniforms.x_shape[${bn}]);
      let W_in = i32(uniforms.x_shape[${yn}]);

      ${r.alignCorners===0?`
      let x_min = -0.5;
      let x_max = f32(W_in) - 0.5;
      let y_min = -0.5;
      let y_max = f32(H_in) - 0.5;
      `:`
      let x_min = 0.0;
      let x_max = f32(W_in) - 1.0;
      let y_min = 0.0;
      let y_max = f32(H_in) - 1.0;
      `};
      let border = vec4<f32>(x_min, y_min, x_max, y_max);

      let indices = ${a.offsetToIndices("global_idx")};
      var grid_indices = vec3<u32>(indices[${Fs}], indices[${bn}], indices[${yn}]);
      let nxy = ${o.getByIndices("grid_indices")};
      var x = gs_denormalize(f32(nxy[0]), W_in);
      var y = gs_denormalize(f32(nxy[1]), H_in);

      ${Qf(a,i,r)}
  }`;return{name:"GridSample",shaderCache:{hint:`${r.cacheKey}`,inputDependencies:["type","type"]},getRunData:d=>{let u=Pe.size(n);return{outputs:[{dims:n,dataType:d[0].dataType}],dispatchGroup:{x:Math.ceil(u/64)},programUniforms:c}},getShaderSource:p}},Jf=(e,r)=>{Uf(e.inputs),e.compute(Xf(e.inputs,r))},Yf=e=>jt({alignCorners:e.align_corners,mode:e.mode,paddingMode:e.padding_mode,format:e.format})}),Kr,Zf,e_,Ul,t_,xo,r_,s_=Ue(()=>{Mt(),Pt(),ur(),ul(),bl(),Ct(),Ys(),Kr=(e,r)=>e.length>r&&e[r].dims.length>0?e[r]:void 0,Zf=(e,r)=>{let t=e[0],s=Kr(e,1),o=Kr(e,2),n=Kr(e,3),a=Kr(e,4),i=Kr(e,5),l=Kr(e,6),c=Kr(e,7);if(t.dims.length!==3&&t.dims.length!==5)throw new Error("Input query is expected to have 3 or 5 dimensions");let p=t.dims[0],d=t.dims[1],u=t.dims.length===3?t.dims[2]:r.numHeads*t.dims[4],_=d,f=0,b=0,A=Math.floor(u/r.numHeads);if(l&&c&&Pe.size(l.dims)&&Pe.size(c.dims)){if(l.dims.length!==4)throw new Error('Input "past_key" is expected to have 4 dimensions');if(l.dims[0]!==p||l.dims[1]!==r.numHeads||l.dims[3]!==A)throw new Error('Input "past_key" shape (batch_size, num_heads, past_sequence_length, head_size)');if(c.dims[0]!==p||c.dims[1]!==r.numHeads||c.dims[3]!==A)throw new Error('Input "past_value" shape (batch_size, num_heads, past_sequence_length, head_size)');if(l.dims[2]!==c.dims[2])throw new Error('Input "past_key" and "past_value" shall have same dim 2 (past_sequence_length)');if(c.dims.length!==4)throw new Error('Input "past_value" is expected to have 4 dimensions');f=l.dims[2],b=l.dims[2]}else if(l&&Pe.size(l.dims)||c&&Pe.size(c.dims))throw new Error('Input "past_key" and "past_value" shall be both present or both absent');let g;if(s&&Pe.size(s.dims)>0){if(t.dims.length!==3)throw new Error('Input "query" is expected to have 3 dimensions when key is given');if(s.dims.length<3||s.dims.length>5)throw new Error('Input "key" is expected to have 3, 4, or 5 dimensions');if(t.dims[0]!==s.dims[0])throw new Error('Input "query" and "key" shall have same dim 0 (batch size)');if(s.dims.length===3){if(s.dims[2]!==t.dims[2])throw new Error('Input "query" and "key" shall have same dim 2 (hidden_size)');g=2,_=s.dims[1]}else if(s.dims.length===5){if(s.dims[2]!==r.numHeads||s.dims[3]!==2||s.dims[4]!==A)throw new Error('Expect "key" shape (batch_size, kv_sequence_length, num_heads, 2, head_size) for packed kv');if(o)throw new Error('Expect "value" be none when "key" has packed kv format.');g=5,_=s.dims[1]}else{if(s.dims[1]!==r.numHeads||s.dims[3]!==A)throw new Error('Expect "key" shape (batch_size, num_heads, kv_sequence_length, head_size) for past_key');g=0,_=s.dims[2]}}else{if(t.dims.length!==5)throw new Error('Input "query" is expected to have 5 dimensions when key is empty');if(t.dims[2]!==r.numHeads||t.dims[3]!==3)throw new Error('Expect "query" shape (batch_size, kv_sequence_length, num_heads, 3, head_size) for packed kv');g=3}if(n&&Pe.size(n.dims)>0){if(n.dims.length!==1)throw new Error('Input "bias" is expected to have 1 dimension');if(s&&s.dims.length===5&&s.dims[3]===2)throw new Error("bias is not allowed for packed kv.")}let y=f+_,C=0;if(a&&Pe.size(a.dims)>0){C=8;let v=a.dims;throw v.length===1?v[0]===p?C=1:v[0]===3*p+2&&(C=3):v.length===2&&v[0]===p&&v[1]===y&&(C=5),C===8?new Error('Input "key_padding_mask" shape shall be (batch_size) or (batch_size, total_sequence_length)'):new Error("Mask not supported")}let x=!1,M=u;if(o&&Pe.size(o.dims)>0){if(o.dims.length!==3&&o.dims.length!==4)throw new Error('Input "value" is expected to have 3 or 4 dimensions');if(t.dims[0]!==o.dims[0])throw new Error('Input "query" and "value" shall have same dim 0 (batch_size)');if(o.dims.length===3){if(_!==o.dims[1])throw new Error('Input "key" and "value" shall have the same dim 1 (kv_sequence_length)');M=o.dims[2]}else{if(_!==o.dims[2])throw new Error('Input "key" and "value" shall have the same dim 2 (kv_sequence_length)');M=o.dims[1]*o.dims[3],x=!0}}let T=!1;if(a&&Pe.size(a.dims)>0)throw new Error("Key padding mask is not supported");if(i&&Pe.size(i.dims)>0){if(i.dims.length!==4)throw new Error('Input "attention_bias" is expected to have 4 dimensions');if(i.dims[0]!==p||i.dims[1]!==r.numHeads||i.dims[2]!==d||i.dims[3]!==y)throw new Error('Expect "attention_bias" shape (batch_size, num_heads, sequence_length, total_sequence_length)')}return{batchSize:p,sequenceLength:d,pastSequenceLength:f,kvSequenceLength:_,totalSequenceLength:y,maxSequenceLength:b,inputHiddenSize:0,hiddenSize:u,vHiddenSize:M,headSize:A,vHeadSize:Math.floor(M/r.numHeads),numHeads:r.numHeads,isUnidirectional:!1,pastPresentShareBuffer:!1,maskFilterValue:r.maskFilterValue,maskType:C,scale:r.scale,broadcastResPosBias:T,passPastInKv:x,qkvFormat:g}},e_=e=>jt({...e}),Ul=jt({perm:[0,2,1,3]}),t_=(e,r,t,s,o,n,a)=>{let i=[s,o,n],l=Pe.size(i),c=[{type:12,data:l},{type:12,data:a},{type:12,data:n}],p=d=>{let u=at("qkv_with_bias",r.dataType,i),_=Oe("qkv",r.dataType,i),f=Oe("bias",t.dataType,i),b=[{name:"output_size",type:"u32"},{name:"bias_offset",type:"u32"},{name:"hidden_size",type:"u32"}];return`
  ${d.registerUniforms(b).declareVariables(_,f,u)}
  ${d.mainStart()}
    ${d.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_size")}
    let bias_offset_idx = (global_idx % uniforms.hidden_size) + uniforms.bias_offset;

    qkv_with_bias[global_idx] = qkv[global_idx] + bias[bias_offset_idx];
  }`};return e.compute({name:"MultiHeadAttentionAddBias",shaderCache:{inputDependencies:["type","type"]},getRunData:()=>({outputs:[{dims:i,dataType:r.dataType,gpuDataType:0}],dispatchGroup:{x:Math.ceil(l/64)},programUniforms:c}),getShaderSource:p},{inputs:[r,t],outputs:[-1]})[0]},xo=(e,r,t,s,o,n,a,i)=>{let l=n;if(a&&Pe.size(a.dims)>0){if(s===1)throw new Error("AddBiasReshape is not implemented. Please export your model with packed QKV or KV");return l=t_(e,n,a,r,s,t*o,i),l=l.reshape([r,s,t,o]),t===1||s===1?l:e.compute(rs(l,Ul.perm),{inputs:[l],outputs:[-1]})[0]}else return n.dims.length===3&&(l=n.reshape([r,s,t,o])),t===1||s===1?l:e.compute(rs(l,Ul.perm),{inputs:[l],outputs:[-1]})[0]},r_=(e,r)=>{let t=Zf(e.inputs,r),s=e.inputs[0],o=Kr(e.inputs,1),n=Kr(e.inputs,2),a=Kr(e.inputs,3),i=Kr(e.inputs,4),l=Kr(e.inputs,5),c=Kr(e.inputs,6),p=Kr(e.inputs,7);if(s.dims.length===5)throw new Error("Packed QKV is not implemented");if((o==null?void 0:o.dims.length)===5)throw new Error("Packed KV is not implemented");let d=o&&n&&o.dims.length===4&&n.dims.length===4,u=xo(e,t.batchSize,t.numHeads,t.sequenceLength,t.headSize,s,a,0);if(d)return wo(e,u,o,n,i,void 0,c,p,l,t);if(!o||!n)throw new Error("key and value must be provided");let _=xo(e,t.batchSize,t.numHeads,t.kvSequenceLength,t.headSize,o,a,t.hiddenSize),f=xo(e,t.batchSize,t.numHeads,t.kvSequenceLength,t.vHeadSize,n,a,2*t.hiddenSize);wo(e,u,_,f,i,void 0,c,p,l,t)}}),n_,o_,a_,i_,Wl,l_,u_,c_=Ue(()=>{Mt(),Pt(),ur(),Ct(),n_=e=>{if(!e||e.length<1)throw new Error("too few inputs")},o_=(e,r)=>{let t=[],s=r.numOutputs;return e[1].dims[0]>0&&(e[1].getBigInt64Array().forEach(o=>t.push(Number(o))),s=t.length),jt({numOutputs:s,axis:r.axis,splitSizes:t})},a_=e=>`
fn calculateOutputIndex(index: u32) -> u32 {
    for (var i: u32 = 0u; i < ${e}u; i += 1u ) {
    if (index < ${it("uniforms.size_in_split_axis","i",e)}) {
        return i;
    }
    }
    return ${e}u;
}`,i_=e=>{let r=e.length,t=[];for(let s=0;s<r;++s){let o=e[s].setByIndices("indices","input[global_idx]");r===1?t.push(o):s===0?t.push(`if (output_number == ${s}u) { ${o} }`):s===r-1?t.push(`else { ${o} }`):t.push(`else if (output_number == ${s}) { ${o} }`)}return`
      fn writeBufferData(output_number: u32, indices: ${e[0].type.indices}, global_idx: u32) {
        ${t.join(`
`)}
      }`},Wl=(e,r)=>{let t=e[0].dims,s=Pe.size(t),o=e[0].dataType,n=Pe.normalizeAxis(r.axis,t.length),a=new Array(r.numOutputs),i=Oe("input",o,t.length),l=new Array(r.numOutputs),c=[],p=[],d=0,u=[{type:12,data:s}];for(let f=0;f<r.numOutputs;f++){d+=r.splitSizes[f],l[f]=d;let b=t.slice();b[n]=r.splitSizes[f],p.push(b),a[f]=at(`output${f}`,o,b.length),c.push({dims:p[f],dataType:e[0].dataType})}u.push({type:12,data:l},...lt(t,...p));let _=f=>`
  ${f.registerUniform("input_size","u32").registerUniform("size_in_split_axis","u32",l.length).declareVariables(i,...a)}
  ${a_(l.length)}
  ${i_(a)}

  ${f.mainStart()}
    ${f.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.input_size")}

    var indices = ${i.offsetToIndices("global_idx")};
    var index = ${i.indicesGet("indices",n)};
    let output_number = calculateOutputIndex(index);
    if (output_number != 0) {
      index -= ${it("uniforms.size_in_split_axis","output_number - 1u",l.length)};
      ${i.indicesSet("indices",n,"index")};
    }
    writeBufferData(output_number, indices, global_idx);
  }`;return{name:"Split",shaderCache:{hint:r.cacheKey,inputDependencies:["rank"]},getShaderSource:_,getRunData:()=>({outputs:c,dispatchGroup:{x:Math.ceil(s/64)},programUniforms:u})}},l_=(e,r)=>{n_(e.inputs);let t=e.inputs.length===1?r:o_(e.inputs,r);e.compute(Wl(e.inputs,t),{inputs:[0]})},u_=e=>{let r=e.axis,t=e.splitSizes,s=e.numOutputs<0?t.length:e.numOutputs;if(s!==t.length)throw new Error("numOutputs and splitSizes lengh must be equal");return jt({axis:r,numOutputs:s,splitSizes:t})}}),d_,ba,p_,h_=Ue(()=>{Mt(),Pt(),ur(),Ct(),d_=(e,r)=>{let[t,s,o,n]=e,{numHeads:a,rotaryEmbeddingDim:i}=r;if(t.dims.length!==3&&t.dims.length!==4)throw new Error(`Input 'x' is expected to have 3 or 4 dimensions, got ${t.dims.length}`);if(!Pe.areEqual(s.dims,[])&&!Pe.areEqual(s.dims,[1])&&s.dims.length!==2)throw new Error(`Input 'position_ids' is expected to have 0, 1, or 2 dimensions, got ${s.dims.length}`);if(o.dims.length!==2)throw new Error(`Input 'cos_cache' is expected to have 2 dimensions, got ${o.dims.length}`);if(n.dims.length!==2)throw new Error(`Input 'sin_cache' is expected to have 2 dimensions, got ${n.dims.length}`);if(!Pe.areEqual(o.dims,n.dims))throw new Error("Inputs 'cos_cache' and 'sin_cache' are expected to have the same shape");if(i>0&&a===0)throw new Error("num_heads must be provided if rotary_embedding_dim is specified");let l=t.dims[0],c=t.dims[t.dims.length-2],p=o.dims[0],d=Pe.sizeFromDimension(t.dims,1)/c,u=i===0?o.dims[1]*2:d/a;if(i>u)throw new Error("rotary_embedding_dim must be less than or equal to head_size");if(s.dims.length===2){if(l!==s.dims[0])throw new Error(`Input 'position_ids' dimension 0 should be of size batch_size, got ${s.dims[0]}`);if(c!==s.dims[1])throw new Error(`Input 'position_ids' dimension 1 should be of size sequence_length, got ${s.dims[1]}`)}if(u/2!==o.dims[1]&&i/2!==o.dims[1])throw new Error(`Input 'cos_cache' dimension 1 should be same as head_size / 2 or rotary_embedding_dim / 2, got ${o.dims[1]}`);if(c>p)throw new Error("Updating cos_cache and sin_cache in RotaryEmbedding is not currently supported")},ba=(e,r)=>{let{interleaved:t,numHeads:s,rotaryEmbeddingDim:o,scale:n}=r,a=e[0].dims[0],i=Pe.sizeFromDimension(e[0].dims,1),l=e[0].dims[e[0].dims.length-2],c=i/l,p=e[2].dims[1],d=o===0?p*2:c/s,u=new Array(a,l,c/d,d-p),_=Pe.computeStrides(u),f=[{type:1,data:n},{type:12,data:u},{type:12,data:_},...e[0].dims.length===3?new Array({type:12,data:[i,c,d,1]}):[],...e[0].dims.length===4?new Array({type:12,data:[i,d,l*d,1]}):[],...lt(e[0].dims,e[1].dims,e[2].dims,e[3].dims,e[0].dims)],b=A=>{let g=Oe("input",e[0].dataType,e[0].dims.length),y=Oe("position_ids",e[1].dataType,e[1].dims.length),C=Oe("cos_cache",e[2].dataType,e[2].dims.length),x=Oe("sin_cache",e[3].dataType,e[3].dims.length),M=at("output",e[0].dataType,e[0].dims.length);return A.registerUniforms([{name:"scale",type:"f32"},{name:"global_shape",type:"u32",length:u.length},{name:"global_strides",type:"u32",length:_.length},{name:"input_output_strides",type:"u32",length:_.length}]),`
        ${A.declareVariables(g,y,C,x,M)}

        ${A.mainStart(Hn)}
          let half_rotary_emb_dim = uniforms.${C.name}_shape[1];
          let bsnh = global_idx / uniforms.global_strides % uniforms.global_shape;
          let size = uniforms.global_shape[0] * uniforms.global_strides[0];
          ${A.guardAgainstOutOfBoundsWorkgroupSizes("size")}

          if (bsnh[3] < half_rotary_emb_dim) {
            let position_ids_idx =
                ${y.broadcastedIndicesToOffset("bsnh.xy",at("",y.type.tensor,2))};
            let position_id =
                u32(${y.getByOffset("position_ids_idx")}) + select(0, bsnh[1], position_ids_idx == 0);
            let i = dot(bsnh, uniforms.input_output_strides) + select(0, bsnh[3], ${t});
            let j = i + select(half_rotary_emb_dim, 1, ${t});
            let re = ${g.getByOffset("i")} * ${C.get("position_id","bsnh[3]")} -
                ${g.getByOffset("j")} * ${x.get("position_id","bsnh[3]")};
            ${M.setByOffset("i","re")}
            let im = ${g.getByOffset("i")} * ${x.get("position_id","bsnh[3]")} +
                ${g.getByOffset("j")} * ${C.get("position_id","bsnh[3]")};
            ${M.setByOffset("j","im")}
          } else {
            let k = dot(bsnh, uniforms.input_output_strides) + half_rotary_emb_dim;
            ${M.setByOffset("k",g.getByOffset("k"))}
          }
        }`};return{name:"RotaryEmbedding",shaderCache:{hint:jt({interleaved:t}).cacheKey,inputDependencies:["rank","rank","rank","rank"]},getShaderSource:b,getRunData:()=>({outputs:[{dims:e[0].dims,dataType:e[0].dataType}],dispatchGroup:{x:Math.ceil(Pe.size(u)/Hn)},programUniforms:f})}},p_=(e,r)=>{d_(e.inputs,r),e.compute(ba(e.inputs,r))}}),m_,f_,Gl,__,g_,MT=Ue(()=>{ur(),Mt(),bl(),s_(),c_(),Ys(),h_(),Ct(),m_=(e,r)=>{if(r.doRotary&&e.length<=7)throw new Error("cos_cache and sin_cache inputs are required if do_rotary is specified");let t=e[0],s=e[1],o=e[2],n=e[3],a=e[4];if(r.doRotary!==0&&e.length<=7)throw new Error("cos_cast and sin_cache are expected if do_rotary attribute is non-zero");if(r.localWindowSize!==-1)throw new Error("Local attention is not supported");if(r.softcap!==0)throw new Error("Softcap is not supported");if(r.rotaryInterleaved!==0)throw new Error("Rotary interleaved is not supported");if(r.smoothSoftmax)throw new Error("Smooth softmax is not supported");if(t.dims.length!==3&&t.dims.length!==5)throw new Error("Input query is expected to have 3 or 5 dimensions");let i=!1,l=t.dims[0],c=t.dims[1],p=t.dims.length===3?i?t.dims[2]/3:t.dims[2]:r.numHeads*t.dims[4],d=c,u=0,_=!s||s.dims.length===0,f=Math.floor(_?p/(r.numHeads+2*r.kvNumHeads):p/r.numHeads);_&&(p=f*r.numHeads);let b=n&&n.dims.length!==0,A=a&&a.dims.length!==0;if(b&&n.dims.length===4&&n.dims[0]===l&&n.dims[1]!==r.kvNumHeads&&n.dims[2]===r.kvNumHeads&&n.dims[3]===f)throw new Error("BSNH pastKey/pastValue is not supported");if(b&&A){if(n.dims.length!==4)throw new Error('Input "past_key" is expected to have 4 dimensions');if(a.dims.length!==4)throw new Error('Input "past_value" is expected to have 4 dimensions');u=n.dims[2]}else if(b||A)throw new Error('Input "past_key" and "past_value" shall be both present or both absent');let g=1;if(s&&s.dims.length>0){if(t.dims.length!==3)throw new Error('Input "query" is expected to have 3 dimensions when key is given');if(s.dims.length<3||s.dims.length>5)throw new Error('Input "key" is expected to have 3, 4, or 5 dimensions');if(t.dims[0]!==s.dims[0])throw new Error('Input "query" and "key" shall have same dim 0 (batch size)');if(s.dims.length===3){if(t.dims[2]%s.dims[2]!==0)throw new Error('Dimension 2 of "query" should be a multiple of "key"');d=s.dims[1]}else if(s.dims.length===5){if(s.dims[2]!==r.numHeads||s.dims[3]!==2||s.dims[4]!==f)throw new Error('Expect "key" shape (batch_size, kv_sequence_length, num_heads, 2, head_size) for packed kv');if(o)throw new Error('Expect "value" be none when "key" has packed kv format.');d=s.dims[1]}else{if(s.dims[1]!==r.numHeads||s.dims[3]!==f)throw new Error('Expect "key" shape (batch_size, num_heads, kv_sequence_length, head_size) for past_key');d=s.dims[2]}}else{if(t.dims.length!==3&&t.dims.length!==5)throw new Error('Input "query" is expected to have 3 or 5 dimensions when key is empty');if(t.dims.length===5&&(t.dims[2]!==r.numHeads||t.dims[3]!==3))throw new Error('Expect "query" shape (batch_size, kv_sequence_length, num_heads, 3, head_size) for packed kv');g=3}let y=0,C=!1,x=r.kvNumHeads?f*r.kvNumHeads:p;if(o&&o.dims.length>0){if(o.dims.length!==3&&o.dims.length!==4)throw new Error('Input "value" is expected to have 3 or 4 dimensions');if(t.dims[0]!==o.dims[0])throw new Error('Input "query" and "value" shall have same dim 0 (batch_size)');if(o.dims.length===3){if(d!==o.dims[1])throw new Error('Input "key" and "value" shall have the same dim 1 (kv_sequence_length)');x=o.dims[2]}else{if(d!==o.dims[2])throw new Error('Input "past_key" and "past_value" shall have the same dim 2 (kv_sequence_length)');x=o.dims[1]*o.dims[3],C=!0}}let M=e.length>4?e[5]:void 0;if(M&&M.dims.length!==1&&M.dims[0]!==l)throw new Error('Input "seqlens" is expected to have 1 dimension and the same dim 0 as batch_size');return{batchSize:l,sequenceLength:c,pastSequenceLength:u,kvSequenceLength:d,totalSequenceLength:-1,maxSequenceLength:-1,inputHiddenSize:0,hiddenSize:p,vHiddenSize:x,headSize:f,vHeadSize:Math.floor(x/r.kvNumHeads),numHeads:r.numHeads,kvNumHeads:r.kvNumHeads,nReps:r.numHeads/r.kvNumHeads,pastPresentShareBuffer:!1,maskType:y,scale:r.scale,broadcastResPosBias:!1,passPastInKv:C,qkvFormat:g}},f_=jt({perm:[0,2,1,3]}),Gl=(e,r,t)=>{let s=r,o=t.kvNumHeads;return r.dims.length===3&&t.kvSequenceLength!==0&&(s=r.reshape([t.batchSize,t.kvSequenceLength,o,t.headSize]),s=e.compute(rs(s,f_.perm),{inputs:[s],outputs:[-1]})[0]),s},__=(e,r,t,s)=>{let o=7,n=["type","type"],a=[e*r],i=e*r,l=[{type:12,data:i},{type:12,data:r},{type:12,data:e}],c=p=>{let d=Oe("seq_lens",t.dataType,t.dims),u=Oe("total_seq_lens",s.dataType,s.dims),_=at("pos_ids",o,a),f=[{name:"output_size",type:"u32"},{name:"sequence_length",type:"u32"},{name:"batch_size",type:"u32"}];return`
  ${p.registerUniforms(f).declareVariables(d,u,_)}
  ${p.mainStart()}
    ${p.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_size")}
    let total_sequence_length = u32(${u.getByOffset("0")});
    let is_subsequent_prompt = uniforms.sequence_length > 1 && uniforms.sequence_length != total_sequence_length;
    let is_first_prompt = !is_subsequent_prompt && uniforms.sequence_length == total_sequence_length;
    let batch_idx = global_idx / uniforms.sequence_length;
    let sequence_idx = i32(global_idx % uniforms.sequence_length);
    var pos_id: i32 = 0;
    let seqlen = ${d.getByOffset("batch_idx")};
    let total_seqlen = seqlen + 1;
    if (is_first_prompt) {
      if (sequence_idx < total_seqlen) {
        pos_id = sequence_idx;
      } else {
        pos_id = 1;
      }
      ${_.setByOffset("global_idx","pos_id")}
    } else if (is_subsequent_prompt) {
      let past_seqlen = total_seqlen - i32(uniforms.sequence_length);
      if (past_seqlen + sequence_idx < total_seqlen) {
        pos_id = past_seqlen + sequence_idx;
      } else {
        pos_id = 1;
      }
      ${_.setByOffset("global_idx","pos_id")}
    } else if (global_idx < uniforms.batch_size) {
      ${_.setByOffset("global_idx","seqlen")}
    };
  }
  `};return{name:"GeneratePositionIds",shaderCache:{hint:`${e};${r}`,inputDependencies:n},getRunData:()=>({outputs:[{dims:a,dataType:o}],dispatchGroup:{x:Math.ceil(i/64)},programUniforms:l}),getShaderSource:c}},g_=(e,r)=>{var x;let t=m_(e.inputs,r);if(e.inputs[0].dims.length===5)throw new Error("Packed QKV is not implemented");if(((x=e.inputs[1])==null?void 0:x.dims.length)===5)throw new Error("Packed KV is not implemented");let s=e.inputs[0],o=e.inputs[1]&&e.inputs[1].dims.length>0?e.inputs[1]:void 0,n=e.inputs[2]&&e.inputs[2].dims.length>0?e.inputs[2]:void 0,a=e.inputs[3]&&e.inputs[3].dims.length!==0?e.inputs[3]:void 0,i=e.inputs[4]&&e.inputs[4].dims.length!==0?e.inputs[4]:void 0,l=e.inputs.length>4?e.inputs[5]:void 0,c=e.inputs.length>5?e.inputs[6]:void 0,p=t.kvNumHeads?t.kvNumHeads:t.numHeads,d=jt({axis:2,numOutputs:3,splitSizes:[t.numHeads*t.headSize,p*t.headSize,p*t.headSize]}),[u,_,f]=!o&&!n?e.compute(Wl([s],d),{inputs:[s],outputs:[-1,-1,-1]}):[s,o,n],b,A;if(r.doRotary){let M=e.compute(__(t.batchSize,t.sequenceLength,l,c),{inputs:[l,c],outputs:[-1]})[0],T=e.inputs[7],v=e.inputs[8],P=jt({interleaved:r.rotaryInterleaved!==0,numHeads:t.numHeads,rotaryEmbeddingDim:0,scale:r.scale}),F=[u,M,T,v],D=[-1];b=e.compute(ba(F,P),{inputs:F,outputs:D})[0],F.splice(0,1,_);let K=jt({interleaved:r.rotaryInterleaved!==0,numHeads:t.kvNumHeads,rotaryEmbeddingDim:0,scale:r.scale});A=e.compute(ba(F,K),{inputs:F,outputs:D})[0]}let g=xo(e,t.batchSize,t.numHeads,t.sequenceLength,t.headSize,r.doRotary?b:u,void 0,0),y=Gl(e,r.doRotary?A:_,t),C=Gl(e,f,t);wo(e,g,y,C,void 0,void 0,a,i,void 0,t,l,c)}}),Hl,w_,M_,b_,bT=Ue(()=>{Mt(),Pt(),Ys(),Ct(),Hl=(e,r,t,s,o,n,a,i)=>{let l=or(n),c=l===1?"f32":`vec${l}f`,p=l===1?"vec2f":`mat2x${l}f`,d=o*a,u=64;d===1&&(u=256);let _=[o,a,n/l],f=[o,a,2],b=["rank","type","type"],A=[];A.push(...lt(_,f));let g=y=>{let C=Oe("x",r.dataType,3,l),x=Oe("scale",t.dataType,t.dims),M=Oe("bias",s.dataType,s.dims),T=at("output",1,3,2),v=[C,x,M,T];return`
  var<workgroup> workgroup_shared : array<${p}, ${u}>;
  const workgroup_size = ${u}u;
  ${y.declareVariables(...v)}
  ${y.mainStart(u)}
    let batch = workgroup_index / uniforms.x_shape[1];
    let channel = workgroup_index % uniforms.x_shape[1];
    let hight = uniforms.x_shape[2];
    // initialize workgroup memory
    var sum = ${c}(0);
    var squared_sum = ${c}(0);
    for (var h = local_idx; h < hight; h += workgroup_size) {
      let value = ${c}(${C.get("batch","channel","h")});
      sum += value;
      squared_sum += value * value;
    }
    workgroup_shared[local_idx] = ${p}(sum, squared_sum);
    workgroupBarrier();

    for (var currSize = workgroup_size >> 1;  currSize > 0; currSize = currSize >> 1) {
      if (local_idx < currSize) {
        workgroup_shared[local_idx] = workgroup_shared[local_idx] + workgroup_shared[local_idx + currSize];
      }
      workgroupBarrier();
    }
    if (local_idx == 0) {
      let sum_final = ${Js("workgroup_shared[0][0]",l)} / f32(hight * ${l});
      let squared_sum_final = ${Js("workgroup_shared[0][1]",l)} / f32(hight * ${l});

      let inv_std_dev = inverseSqrt(squared_sum_final - sum_final * sum_final + f32(${i}));
      let channel_scale = inv_std_dev * f32(scale[channel]);
      let channel_shift = f32(bias[channel]) - sum_final * channel_scale;
      output[workgroup_index] = vec2f(channel_scale, channel_shift);
    }
  }`};return e.compute({name:"InstanceNormComputeChannelScaleShift",shaderCache:{hint:`${l};${i};${u}`,inputDependencies:b},getRunData:()=>({outputs:[{dims:f,dataType:1}],dispatchGroup:{x:d},programUniforms:A}),getShaderSource:g},{inputs:[r,t,s],outputs:[-1]})[0]},w_=(e,r,t)=>{let s=r[0].dims,o=s,n=2,a=s[0],i=s[1],l=Pe.sizeFromDimension(s,n),c=or(l),p=Pe.size(o)/c,d=Hl(e,r[0],r[1],r[2],a,l,i,t.epsilon),u=[a,i,l/c],_=[a,i],f=["type","none"],b=A=>{let g=Oe("x",r[0].dataType,u.length,c),y=Oe("scale_shift",1,_.length,2),C=at("output",r[0].dataType,u.length,c),x=[g,y,C];return`
  ${A.registerUniform("output_size","u32").declareVariables(...x)}
  ${A.mainStart()}
  ${A.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_size")}
      let outputIndices = ${C.offsetToIndices("global_idx")};
      let batch = outputIndices[0];
      let channel = outputIndices[1];
      let scale_shift = ${y.getByIndices("vec2<u32>(batch, channel)")};
      let value = ${g.getByOffset("global_idx")} * ${C.type.value}(scale_shift.x) + ${C.type.value}(scale_shift.y);
      ${C.setByOffset("global_idx","value")};
  }`};e.compute({name:"InstanceNormalization",shaderCache:{hint:`${c}`,inputDependencies:f},getRunData:()=>({outputs:[{dims:o,dataType:r[0].dataType}],dispatchGroup:{x:Math.ceil(p/64)},programUniforms:[{type:12,data:p},...lt(u,_,u)]}),getShaderSource:b},{inputs:[r[0],d]})},M_=(e,r,t)=>{let s=r[0].dims,o=s,n=s[0],a=s[s.length-1],i=Pe.sizeFromDimension(s,1)/a,l=or(a),c=Pe.size(o)/l,p=[{type:12,data:i},{type:12,data:Math.floor(a/l)}],d=["type","type"],u=!1,_=[0,s.length-1];for(let g=0;g<s.length-2;g++)u=u||s[g+1]!==1,_.push(g+1);u=u&&s[s.length-1]!==1;let f=u?e.compute(rs(e.inputs[0],_),{inputs:[e.inputs[0]],outputs:[-1]})[0]:e.inputs[0].reshape(Array.from({length:s.length},(g,y)=>s[_[y]])),b=Hl(e,f,r[1],r[2],n,i,a,t.epsilon),A=g=>{let y=Sr(r[0].dataType),C=l===1?"vec2f":`mat${l}x2f`,x=v=>{let P=v===0?"x":"y",F=l===1?"f32":`vec${l}f`;switch(l){case 1:return`${y}(${F}(scale.${P}))`;case 2:return`vec2<${y}>(${F}(scale[0].${P}, scale[1].${P}))`;case 4:return`vec4<${y}>(${F}(scale[0].${P}, scale[1].${P}, scale[2].${P}, scale[3].${P}))`;default:throw new Error(`Not supported compoents ${l}`)}},M=Oe("input",r[0].dataType,r[0].dims,l),T=at("output",r[0].dataType,o,l);return`
  @group(0) @binding(0) var<storage, read> input : array<${M.type.storage}>;
  @group(0) @binding(1) var<storage, read> scale_input : array<${C}>;
  @group(0) @binding(2) var<storage, read_write> output : array<${T.type.storage}>;
  struct Uniforms {H: u32, C : u32};
  @group(0) @binding(3) var<uniform> uniforms: Uniforms;

  ${g.mainStart()}
    let current_image_number = global_idx / (uniforms.C * uniforms.H);
    let current_channel_number = global_idx % uniforms.C;

    let scale_offset = current_image_number * uniforms.C + current_channel_number;
    let scale = scale_input[scale_offset];
    output[global_idx] = fma(input[global_idx], ${x(0)}, ${x(1)});
  }`};e.compute({name:"InstanceNormalizationNHWC",shaderCache:{hint:`${l}`,inputDependencies:d},getRunData:()=>({outputs:[{dims:o,dataType:r[0].dataType}],dispatchGroup:{x:Math.ceil(c/64)},programUniforms:p}),getShaderSource:A},{inputs:[r[0],b]})},b_=(e,r)=>{r.format==="NHWC"?M_(e,e.inputs,r):w_(e,e.inputs,r)}}),y_,v_,x_,yT=Ue(()=>{Mt(),Pt(),Ct(),y_=e=>{if(!e||e.length<2)throw new Error("layerNorm requires at least 2 inputs.")},v_=(e,r,t)=>{let s=r.simplified,o=e[0].dims,n=e[1],a=!s&&e[2],i=o,l=Pe.normalizeAxis(r.axis,o.length),c=Pe.sizeToDimension(o,l),p=Pe.sizeFromDimension(o,l),d=Pe.size(n.dims),u=a?Pe.size(a.dims):0;if(d!==p||a&&u!==p)throw new Error(`Size of X.shape()[axis:] == ${p}.
       Size of scale and bias (if provided) must match this.
       Got scale size of ${d} and bias size of ${u}`);let _=[];for(let M=0;M<o.length;++M)M<l?_.push(o[M]):_.push(1);let f=or(p),b=["type","type"],A=[{type:12,data:c},{type:1,data:p},{type:12,data:Math.floor(p/f)},{type:1,data:r.epsilon}];a&&b.push("type");let g=t>1,y=t>2,C=M=>{let T=Sr(e[0].dataType),v=[Oe("x",e[0].dataType,e[0].dims,f),Oe("scale",n.dataType,n.dims,f)];a&&v.push(Oe("bias",a.dataType,a.dims,f)),v.push(at("output",e[0].dataType,i,f)),g&&v.push(at("mean_data_output",1,_)),y&&v.push(at("inv_std_output",1,_));let P=[{name:"norm_count",type:"u32"},{name:"norm_size",type:"f32"},{name:"norm_size_vectorized",type:"u32"},{name:"epsilon",type:"f32"}];return`
  ${M.registerUniforms(P).declareVariables(...v)}
  ${M.mainStart()}
    ${M.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.norm_count")}
    let offset = global_idx * uniforms.norm_size_vectorized;
    var mean_vector = ${hl("f32",f)};
    var mean_square_vector = ${hl("f32",f)};

    for (var h: u32 = 0u; h < uniforms.norm_size_vectorized; h++) {
      let value = ${Kn(T,f,"x[h + offset]")};
      mean_vector += value;
      mean_square_vector += value * value;
    }
    let mean = ${Js("mean_vector",f)} / uniforms.norm_size;
    let inv_std_dev = inverseSqrt(${Js("mean_square_vector",f)} / uniforms.norm_size ${s?"":"- mean * mean"} + uniforms.epsilon);

    for (var j: u32 = 0; j < uniforms.norm_size_vectorized; j++) {
      let f32input = ${Kn(T,f,"x[j + offset]")};
      let f32scale = ${Kn(T,f,"scale[j]")};
      output[j + offset] = ${v[0].type.value}((f32input ${s?"":"- mean"}) * inv_std_dev * f32scale
        ${a?`+ ${Kn(T,f,"bias[j]")}`:""}
      );
    }

    ${g?"mean_data_output[global_idx] = mean":""};
    ${y?"inv_std_output[global_idx] = inv_std_dev":""};
  }`},x=[{dims:i,dataType:e[0].dataType}];return g&&x.push({dims:_,dataType:1}),y&&x.push({dims:_,dataType:1}),{name:"LayerNormalization",shaderCache:{hint:`${f};${t};${s}`,inputDependencies:b},getRunData:()=>({outputs:x,dispatchGroup:{x:Math.ceil(c/64)},programUniforms:A}),getShaderSource:C}},x_=(e,r)=>{y_(e.inputs),e.compute(v_(e.inputs,r,e.outputCount))}}),T_,E_,vT=Ue(()=>{Pt(),Sl(),Il(),T_=e=>{if(!e||e.length!==2)throw new Error("MatMul requires 2 inputs.");if(e[0].dims[e[0].dims.length-1]!==e[1].dims[e[1].dims.length-2])throw new Error("shared dimension does not match.")},E_=e=>{T_(e.inputs);let r=Gn.calcShape(e.inputs[0].dims,e.inputs[1].dims,!0);if(!r)throw new Error("Can't use matmul on the given tensors");let t=r[r.length-1],s=e.inputs[0].dims[e.inputs[0].dims.length-1];if(t<8&&s<8)e.compute(Cl(e.inputs,{activation:""},r));else{let o=r[r.length-2],n=Pe.size(e.inputs[0].dims.slice(0,-2)),a=Pe.size(e.inputs[1].dims.slice(0,-2));if(n!==1&&o===1&&a===1){let i=e.inputs[0].reshape([1,n,s]),l=e.inputs[1].reshape([1,s,t]),c=[1,n,t],p=[i,l];e.compute(_a(p,{activation:""},r,c),{inputs:p})}else e.compute(_a(e.inputs,{activation:""},r))}}}),P_,C_,S_,$_,A_,xT=Ue(()=>{Mt(),Pt(),ur(),Ct(),P_=(e,r)=>{if(e.length<3||e.length>4)throw new Error("MatMulNBits requires 3 or 4 inputs");let t=e[0],s=t.dims.length;if(t.dims[s-1]!==r.k)throw new Error("The last dim of input shape does not match the k value");let o=Math.floor((r.k+r.blockSize-1)/r.blockSize),n=r.blockSize/8*r.bits,a=e[1];if(!Pe.areEqual(a.dims,[r.n,o,n]))throw new Error("The second inputs must be 3D tensor with shape N X nBlocksPerCol X blobSize");let i=e[2].dims;if(Pe.size(i)!==r.n*o)throw new Error("scales input size error.");if(e.length===4){let l=e[3].dims,c=r.bits>4?r.n*o:r.n*Math.floor((o+1)/2);if(Pe.size(l)!==c)throw new Error("zeroPoints input size error.")}},C_=(e,r)=>{let t=e[0].dims,s=t.length,o=t[s-2],n=r.k,a=r.n,i=t.slice(0,s-2),l=Pe.size(i),c=e[1].dims[2]/4,p=e[0].dataType,d=or(r.k),u=or(c),_=or(a),f=i.concat([o,a]),b=o>1&&a/_%2===0?2:1,A=Pe.size(f)/_/b,g=64,y=[],C=[l,o,n/d],x=Pe.convertShape(e[1].dims).slice();x.splice(-1,1,c/u),y.push(...lt(C)),y.push(...lt(x)),y.push(...lt(e[2].dims)),e.length===4&&y.push(...lt(Pe.convertShape(e[3].dims)));let M=[l,o,a/_];y.push(...lt(M));let T=v=>{let P=C.length,F=Oe("a",e[0].dataType,P,d),D=Oe("b",12,x.length,u),K=Oe("scales",e[2].dataType,e[2].dims.length),U=[F,D,K],j=e.length===4?Oe("zero_points",12,e[3].dims.length):void 0;j&&U.push(j);let ne=M.length,q=at("output",e[0].dataType,ne,_),te=Sr(e[0].dataType),Z=(()=>{switch(d){case 1:return`array<${te}, 8>`;case 2:return`mat4x2<${te}>`;case 4:return`mat2x4<${te}>`;default:throw new Error(`${d}-component is not supported.`)}})(),ae=()=>{let B=`
          // reuse a data
            var input_offset = ${F.indicesToOffset(`${F.type.indices}(batch, row, word_offset)`)};
            var a_data: ${Z};
            for (var j: u32 = 0; j < ${8/d}; j++) {
              a_data[j] = ${F.getByOffset("input_offset")};
              input_offset++;
            }
          `;for(let O=0;O<_*b;O++)B+=`
            b_value = ${u===1?`b${O}_data`:`b${O}_data[i]`};
            b_value_lower = unpack4xU8(b_value & b_mask);
            b_value_upper = unpack4xU8((b_value >> 4) & b_mask);
            b_quantized_values = ${Z}(${Array.from({length:4},(W,N)=>`${te}(b_value_lower[${N}]), ${te}(b_value_upper[${N}])`).join(", ")});
            b_dequantized_values = ${d===1?`${Z}(${Array.from({length:8},(W,N)=>`(b_quantized_values[${N}] - ${j?`zero_point${O}`:"zero_point"}) * scale${O}`).join(", ")});`:`(b_quantized_values - ${Z}(${Array(8).fill(`${j?`zero_point${O}`:"zero_point"}`).join(",")})) * scale${O};`};
            workgroup_shared[local_id.x * ${b} + ${Math.floor(O/_)}]${_>1?`[${O%_}]`:""} += ${Array.from({length:8/d},(W,N)=>`${d===1?`a_data[${N}] * b_dequantized_values[${N}]`:`dot(a_data[${N}], b_dequantized_values[${N}])`}`).join(" + ")};
          `;return B},he=()=>{let B=`
            var col_index = col * ${_};
            ${j?`
            let zero_point_bytes_per_col = (nBlocksPerCol + 1) / 2;
            var zero_point_byte_count: u32;
            var zero_point_word_index: u32;
            var zero_point_byte_offset: u32;
            let zero_point_nibble_offset: u32 = block & 0x1u;
            var zero_point_bits_offset: u32;
            var zero_point_word: u32;`:`
            // The default zero point is 8 for unsigned 4-bit quantization.
            let zero_point = ${te}(8);`}
            `;for(let O=0;O<_*b;O++)B+=`
            let scale${O} = ${K.getByOffset("col_index * nBlocksPerCol + block")};
            ${j?`
            zero_point_byte_count = col_index * zero_point_bytes_per_col + (block >> 0x1u);
            zero_point_word_index = zero_point_byte_count >> 0x2u;
            zero_point_byte_offset = zero_point_byte_count & 0x3u;
            zero_point_bits_offset = (zero_point_byte_offset << 3) + (zero_point_nibble_offset << 2);
            zero_point_word = ${j.getByOffset("zero_point_word_index")} >> zero_point_bits_offset;
            let zero_point${O} = ${te}((zero_point_word) & 0xFu);`:""}
            col_index += 1;`;return B},Q=()=>{let B=`col_index = col * ${_};`;for(let O=0;O<_*b;O++)B+=`
            let b${O}_data = ${D.getByIndices(`${D.type.indices}(col_index, block, word)`)};
            col_index += 1;`;return B+=`
            var b_value: u32;
            let b_mask: u32 = 0x0F0F0F0Fu;
            var b_value_lower: vec4<u32>;
            var b_value_upper: vec4<u32>;
            var b_quantized_values: ${Z};
            var b_dequantized_values: ${Z};`,B};return`
        var<workgroup> workgroup_shared: array<${q.type.value}, ${b*g}>;
        ${v.declareVariables(...U,q)}
        ${v.mainStart([g,1,1])}
          let output_indices = ${q.offsetToIndices(`(global_idx / ${g}) * ${b}`)};
          let col = output_indices[2];
          let row = output_indices[1];
          let batch = output_indices[0];
          let nBlocksPerCol = uniforms.b_shape[1];

          for (var block = local_id.x; block < nBlocksPerCol; block += ${g}) {
            //process one block
            var word_offset: u32 = block * ${r.blockSize/d};
            ${he()}
            for (var word: u32 = 0; word < ${c}; word += ${u}) {
              ${Q()}
              for (var i: u32 = 0; i < ${u}; i++) {
                ${ae()}
                word_offset += ${8/d};
              }
            }
          }
          workgroupBarrier();

          if (local_id.x < ${b}) {
            var output_value: ${q.type.value} = ${q.type.value}(0);
            var workgroup_shared_offset: u32 = local_id.x;
            for (var b: u32 = 0u; b < ${g}u; b++) {
              output_value += workgroup_shared[workgroup_shared_offset];
              workgroup_shared_offset += ${b};
            }
            ${q.setByIndices(`${q.type.indices}(batch, row, col + local_id.x)`,"output_value")};
          }
        }`};return{name:"MatMulNBits",shaderCache:{hint:`${r.blockSize};${r.bits};${d};${u};${_};${b};${g}`,inputDependencies:Array(e.length).fill("rank")},getRunData:()=>({outputs:[{dims:f,dataType:p}],dispatchGroup:{x:A},programUniforms:y}),getShaderSource:T}},S_=(e,r)=>{let t=e[0].dims,s=t.length,o=t[s-2],n=r.k,a=r.n,i=t.slice(0,s-2),l=Pe.size(i),c=e[1].dims[2]/4,p=e[0].dataType,d=or(r.k),u=or(c),_=i.concat([o,a]),f=128,b=a%8===0?8:a%4===0?4:1,A=f/b,g=A*u*8,y=g/d,C=g/r.blockSize,x=Pe.size(_)/b,M=[],T=[l,o,n/d],v=Pe.convertShape(e[1].dims).slice();v.splice(-1,1,c/u),M.push(...lt(T)),M.push(...lt(v)),M.push(...lt(e[2].dims)),e.length===4&&M.push(...lt(Pe.convertShape(e[3].dims)));let P=[l,o,a];M.push(...lt(P));let F=D=>{let K=T.length,U=Oe("a",e[0].dataType,K,d),j=Oe("b",12,v.length,u),ne=Oe("scales",e[2].dataType,e[2].dims.length),q=[U,j,ne],te=e.length===4?Oe("zero_points",12,e[3].dims.length):void 0;te&&q.push(te);let Z=P.length,ae=at("output",e[0].dataType,Z),he=Sr(e[0].dataType),Q=()=>{switch(d){case 1:return`
          let a_data0 = vec4<${he}>(sub_a[word_offset], sub_a[word_offset + 1], sub_a[word_offset + 2], sub_a[word_offset + 3]);
          let a_data1 = vec4<${he}>(sub_a[word_offset + 4], sub_a[word_offset + 5], sub_a[word_offset + 6], sub_a[word_offset + 7]);`;case 2:return`
          let a_data0 = vec4<${he}>(sub_a[word_offset], sub_a[word_offset + 1]);
          let a_data1 = vec4<${he}>(sub_a[word_offset + 2], sub_a[word_offset + 3]);`;case 4:return`
          let a_data0 = sub_a[word_offset];
          let a_data1 = sub_a[word_offset + 1];`;default:throw new Error(`${d}-component is not supported.`)}};return`
        var<workgroup> sub_a: array<${U.type.value}, ${y}>;
        var<workgroup> inter_results: array<array<${ae.type.value}, ${A}>, ${b}>;
        ${D.declareVariables(...q,ae)}
        ${D.mainStart([A,b,1])}
          let output_indices = ${ae.offsetToIndices(`workgroup_index * ${b}`)};
          let col = output_indices[2];
          let row = output_indices[1];
          let batch = output_indices[0];
          let n_blocks_per_col = uniforms.b_shape[1];
          let num_tiles =  (n_blocks_per_col - 1) / ${C} + 1;

          // Loop over shared dimension.
          for (var tile: u32 = 0; tile < num_tiles; tile += 1) {
            let a_col_start = tile * ${y};
            // load one tile A data into shared memory.
            for (var a_offset = local_idx; a_offset < ${y}; a_offset += ${f})
            {
              let a_col = a_col_start + a_offset;
              if (a_col < uniforms.a_shape[2])
              {
                sub_a[a_offset] = ${U.getByIndices(`${U.type.indices}(batch, row, a_col)`)};
              } else {
                sub_a[a_offset] = ${U.type.value}(0);
              }
            }
            workgroupBarrier();

            // each thread process one block
            let b_row = col + local_id.y;
            let block = tile * ${C} + local_id.x;
            ${te?`
            let zero_point_bytes_per_col = (n_blocks_per_col + 1) / 2;
            let zero_point_byte_count = b_row * zero_point_bytes_per_col + (block >> 0x1u);
            let zero_point_word_index = zero_point_byte_count >> 0x2u;
            let zero_point_byte_offset = zero_point_byte_count & 0x3u;
            let zero_point_nibble_offset: u32 = block & 0x1u;
            let zero_point_bits_offset = (zero_point_byte_offset << 3) + (zero_point_nibble_offset << 2);
            let zero_point_word = ${te.getByOffset("zero_point_word_index")} >> zero_point_bits_offset;
            let zero_point = ${he}((zero_point_word) & 0xFu);`:`
            // The default zero point is 8 for unsigned 4-bit quantization.
            let zero_point = ${he}(8);`}
            let scale = ${ne.getByOffset("b_row * n_blocks_per_col + block")};
            let b_data = ${j.getByIndices(`${j.type.indices}(b_row, block, 0)`)};
            var word_offset = local_id.x * ${r.blockSize/d};
            for (var i: u32 = 0; i < ${u}; i++) {
              ${Q()}
              let b_value = ${u===1?"b_data":"b_data[i]"};
              let b_value_lower = unpack4xU8(b_value & 0x0F0F0F0Fu);
              let b_value_upper = unpack4xU8((b_value >> 4) & 0x0F0F0F0Fu);
              let b_quantized_values = mat2x4<${he}>(${Array.from({length:4},(B,O)=>`${he}(b_value_lower[${O}]), ${he}(b_value_upper[${O}])`).join(", ")});
              let b_dequantized_values = (b_quantized_values - mat2x4<${he}>(${Array(8).fill("zero_point").join(",")})) * scale;
              inter_results[local_id.y][local_id.x] += ${Array.from({length:2},(B,O)=>`${`dot(a_data${O}, b_dequantized_values[${O}])`}`).join(" + ")};
              word_offset += ${8/d};
            }
            workgroupBarrier();
          }

          if (local_idx < ${b}) {
            var output_value: ${ae.type.value} = ${ae.type.value}(0);
            for (var b = 0u; b < ${A}; b++) {
              output_value += inter_results[local_idx][b];
            }
            if (col + local_idx < uniforms.output_shape[2])
            {
              ${ae.setByIndices(`${ae.type.indices}(batch, row, col + local_idx)`,"output_value")}
            }
          }
        }`};return{name:"BlockwiseMatMulNBits32",shaderCache:{hint:`${r.blockSize};${d};${u};${A};${b}`,inputDependencies:Array(e.length).fill("rank")},getRunData:()=>({outputs:[{dims:_,dataType:p}],dispatchGroup:{x},programUniforms:M}),getShaderSource:F}},$_=(e,r)=>{P_(e.inputs,r),r.blockSize===32&&e.adapterInfo.isVendor("intel")&&e.adapterInfo.isArchitecture("gen-12lp")?e.compute(S_(e.inputs,r)):e.compute(C_(e.inputs,r))},A_=e=>jt(e)}),k_,I_,F_,O_,D_,L_,z_,R_,B_,TT=Ue(()=>{Mt(),Pt(),Ct(),k_=e=>{if(!e||e.length<1)throw new Error("Too few inputs");if(e[0].dataType!==1&&e[0].dataType!==10)throw new Error("Input type must be float or float16.");if(e.length>=2){let r=e[0].dims.length*2===e[1].dims[0];if(e.length===4&&(r=e[3].dims[0]*2===e[1].dims[0]),!r)throw new Error("The pads should be a 1D tensor of shape [2 * input_rank] or [2 * num_axes].")}},I_=(e,r,t)=>{let s="";for(let o=r-1;o>=0;--o)s+=`
            k = i32(${e.indicesGet("indices",o)}) - ${it("uniforms.pads",o,t)};
            if (k < 0) {
              break;
            }
            if (k >= i32(${it("uniforms.x_shape",o,r)})) {
              break;
            }
            offset += k * i32(${it("uniforms.x_strides",o,r)});
        `;return`
          value = ${e.type.value}(uniforms.constant_value);
          for (var i = 0; i < 1; i++) {
            var offset = 0;
            var k = 0;
            ${s}
            value = x[offset];
          }
      `},F_=(e,r,t)=>{let s="";for(let o=r-1;o>=0;--o)s+=`
                k = i32(${e.indicesGet("indices",o)}) - ${it("uniforms.pads",o,t)};
                if (k < 0) {
                  k = -k;
                }
                {
                  let _2n_1 = 2 * (i32(${it("uniforms.x_shape",o,r)}) - 1);
                  k = k % _2n_1;
                  if(k >= i32(${it("uniforms.x_shape",o,r)})) {
                    k = _2n_1 - k;
                  }
                }
                offset += k * i32(${it("uniforms.x_strides",o,r)});
            `;return`
              var offset = 0;
              var k = 0;
              ${s}
              value = x[offset];
          `},O_=(e,r,t)=>{let s="";for(let o=r-1;o>=0;--o)s+=`
                k = i32(${e.indicesGet("indices",o)}) - ${it("uniforms.pads",o,t)};
                if (k < 0) {
                  k = 0;
                }
                if (k >= i32(${it("uniforms.x_shape",o,r)})) {
                  k = i32(${it("uniforms.x_shape",o,r)}) - 1;
                }
                offset += k * i32(${it("uniforms.x_strides",o,r)});
            `;return`
              var offset = 0;
              var k = 0;
              ${s}
              value = x[offset];
          `},D_=(e,r,t)=>{let s="";for(let o=r-1;o>=0;--o)s+=`
                k = i32(${e.indicesGet("indices",o)}) - ${it("uniforms.pads",o,t)};
                if (k < 0)  {
                  k += i32(${it("uniforms.x_shape",o,r)}]);
                }
                if (k >= i32(${it("uniforms.x_shape",o,r)})) {
                  k -= i32(${it("uniforms.x_shape",o,r)});
                }
                offset += k * i32(${it("uniforms.x_strides",o,r)});
            `;return`
              var offset = 0;
              var k = 0;
              ${s}
              value = x[offset];
          `},L_=(e,r,t)=>{switch(t.mode){case 0:return I_(e,r,t.pads.length);case 1:return F_(e,r,t.pads.length);case 2:return O_(e,r,t.pads.length);case 3:return D_(e,r,t.pads.length);default:throw new Error("Invalid mode")}},z_=(e,r)=>{let t=Pe.padShape(e[0].dims.slice(),r.pads),s=e[0].dims,o=Pe.size(t),n=[{type:12,data:o},{type:6,data:r.pads}],a=e.length>=3&&e[2].data;r.mode===0&&n.push({type:a?e[2].dataType:1,data:r.value}),n.push(...lt(e[0].dims,t));let i=["rank"],l=c=>{let p=at("output",e[0].dataType,t.length),d=Oe("x",e[0].dataType,s.length),u=d.type.value,_=L_(p,s.length,r),f=[{name:"output_size",type:"u32"},{name:"pads",type:"i32",length:r.pads.length}];return r.mode===0&&f.push({name:"constant_value",type:a?u:"f32"}),`
            ${c.registerUniforms(f).declareVariables(d,p)}
            ${c.mainStart()}
            ${c.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_size")}

            let indices = ${p.offsetToIndices("global_idx")};

            var value = ${u}(0);
            ${_}
            output[global_idx] = value;
        }`};return{name:"Pad",shaderCache:{hint:`${r.mode}${a}`,inputDependencies:i},getRunData:()=>({outputs:[{dims:t,dataType:e[0].dataType}],dispatchGroup:{x:Math.ceil(Pe.size(t)/64)},programUniforms:n}),getShaderSource:l}},R_=(e,r)=>{if(e.length>1){let t=e[1].getBigInt64Array(),s=e.length>=3&&e[2].data?e[2].dataType===10?e[2].getUint16Array()[0]:e[2].getFloat32Array()[0]:0,o=e[0].dims.length,n=new Int32Array(2*o).fill(0);if(e.length>=4){let i=e[3].getBigInt64Array();for(let l=0;l<i.length;l++)n[Number(i[l])]=Number(t[l]),n[Number(i[l])+o]=Number(t[l+i.length])}else t.forEach((i,l)=>n[Number(l)]=Number(i));let a=[];return n.forEach(i=>a.push(i)),{mode:r.mode,value:s,pads:a}}else return r},B_=(e,r)=>{k_(e.inputs);let t=R_(e.inputs,r);e.compute(z_(e.inputs,t),{inputs:[0]})}}),To,Kl,ql,Ql,Xl,N_,j_,Jl,Yl,V_,U_,Zl,W_,G_,eu,H_,K_,q_,Q_,ET=Ue(()=>{bs(),Mt(),Pt(),Ct(),To=e=>{if(Jt.webgpu.validateInputContent&&(!e||e.length!==1))throw new Error("Pool ops requires 1 input.")},Kl=(e,r,t)=>{let s=r.format==="NHWC",o=e.dims.slice();s&&o.splice(1,0,o.pop());let n=Object.hasOwnProperty.call(r,"dilations"),a=r.kernelShape.slice(),i=r.strides.slice(),l=n?r.dilations.slice():[],c=r.pads.slice();la.adjustPoolAttributes(t,o,a,i,l,c);let p=la.computePoolOutputShape(t,o,i,l,a,c,r.autoPad),d=Object.assign({},r);n?Object.assign(d,{kernelShape:a,strides:i,pads:c,dilations:l,cacheKey:r.cacheKey}):Object.assign(d,{kernelShape:a,strides:i,pads:c,cacheKey:r.cacheKey});let u=p.slice();return u.push(u.splice(1,1)[0]),[d,s?u:p]},ql=(e,r)=>{let t=r.format==="NHWC",s=Pe.size(e),o=Pe.size(r.kernelShape),n=[{type:12,data:s},{type:12,data:o}],a=[{name:"outputSize",type:"u32"},{name:"kernelSize",type:"u32"}];if(r.kernelShape.length<=2){let i=r.kernelShape[r.kernelShape.length-1],l=r.strides[r.strides.length-1],c=r.pads[r.pads.length/2-1],p=r.pads[r.pads.length-1],d=!!(c+p);n.push({type:12,data:i},{type:12,data:l},{type:12,data:c},{type:12,data:p}),a.push({name:"kw",type:"u32"},{name:"sw",type:"u32"},{name:"pwStart",type:"u32"},{name:"pwEnd",type:"u32"});let u=!1;if(r.kernelShape.length===2){let _=r.kernelShape[r.kernelShape.length-2],f=r.strides[r.strides.length-2],b=r.pads[r.pads.length/2-2],A=r.pads[r.pads.length-2];u=!!(b+A),n.push({type:12,data:_},{type:12,data:f},{type:12,data:b},{type:12,data:A}),a.push({name:"kh",type:"u32"},{name:"sh",type:"u32"},{name:"phStart",type:"u32"},{name:"phEnd",type:"u32"})}return[n,a,!0,d,u]}else{if(t)throw new Error("Pooling with kernelShape.length > 2 is not supported for NHWC format.");let i=Pe.computeStrides(r.kernelShape);n.push({type:12,data:i},{type:12,data:r.pads},{type:12,data:r.strides}),a.push({name:"kernelStrides",type:"u32",length:i.length},{name:"pads",type:"u32",length:r.pads.length},{name:"strides",type:"u32",length:r.strides.length});let l=r.pads.reduce((c,p)=>c+p);return[n,a,!!l,!1,!1]}},Ql=(e,r,t,s,o,n,a,i,l,c,p,d)=>{let u=o.format==="NHWC",_=r.type.value,f=at("output",r.type.tensor,s);if(o.kernelShape.length<=2){let b="",A="",g="",y=t-(u?2:1);if(p?b=`
                for (var i: u32 = 0u; i < uniforms.kw; i++) {
                  xIndices[${y}] = indices[${y}] * uniforms.sw - uniforms.pwStart + i;
                  if (xIndices[${y}] < 0 || xIndices[${y}]
                      >= uniforms.x_shape[${y}]) {
                    pad++;
                    continue;
                  }
                  let x_val = x[${r.indicesToOffset("xIndices")}];
                  ${n}
                }`:b=`
                for (var i: u32 = 0u; i < uniforms.kw; i++) {
                  xIndices[${y}] = indices[${y}] * uniforms.sw - uniforms.pwStart + i;
                  let x_val = x[${r.indicesToOffset("xIndices")}];
                  ${n}
                }`,o.kernelShape.length===2){let C=t-(u?3:2);d?A=`
                for (var j: u32 = 0u; j < uniforms.kh; j++) {
                  xIndices[${C}] = indices[${C}] * uniforms.sh - uniforms.phStart + j;
                  if (xIndices[${C}] < 0 || xIndices[${C}] >= uniforms.x_shape[${C}]) {
                    pad += i32(uniforms.kw);
                    continue;
                  }
              `:A=`
                for (var j: u32 = 0u; j < uniforms.kh; j++) {
                  xIndices[${C}] = indices[${C}] * uniforms.sh - uniforms.phStart + j;
                `,g=`
              }
            `}return`
            ${e.registerUniforms(l).declareVariables(r,f)}

            ${e.mainStart()}
              ${e.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.outputSize")}

              let indices = ${f.offsetToIndices("global_idx")};
              var xIndices = ${f.offsetToIndices("global_idx")};

              var value = ${_}(${i});
              var pad = 0;
              ${A}
              ${b}
              ${g}
              ${a}

              output[global_idx] = value;
            }`}else{if(u)throw new Error("Pooling with kernelShape.length > 2 is not supported for NHWC format.");let b=o.kernelShape.length,A=o.pads.length,g="";return c?g=`
                if (xIndices[j] >= uniforms.x_shape[j]) {
                  pad++;
                  isPad = true;
                  break;
                }
              }
              if (!isPad) {
                let x_val = x[${r.indicesToOffset("xIndices")}];
                ${n}
              }`:g=`
              }
              let x_val = x[${r.indicesToOffset("xIndices")}];
              ${n}
            `,`
            ${e.registerUniforms(l).declareVariables(r,f)}

            ${e.mainStart()}
              ${e.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.outputSize")}
              let indices = ${f.offsetToIndices("global_idx")};
              var xIndices = ${f.offsetToIndices("global_idx")};

              var offsets: array<u32, ${b}>;

              var value = ${_}(${i});
              var pad = 0;
              var isPad = false;

              for (var i: u32 = 0u; i < uniforms.kernelSize; i++) {
                var offset = i;
                for (var j = 0u; j < ${b-1}u; j++) {
                  offsets[j] = offset / ${it("uniforms.kernelStrides","j",b)};
                  offset -= offsets[j] * ${it("uniforms.kernelStrides","j",b)};
                }
                offsets[${b-1}] = offset;

                isPad = false;
                for (var j = ${t-b}u; j < ${t}u; j++) {
                  xIndices[j] = indices[j] * ${it("uniforms.strides",`j - ${t-b}u`,b)}
                    + offsets[j - ${t-b}u] - ${it("uniforms.pads","j - 2u",A)};
                  ${g}
              }
              ${a}

              output[global_idx] = value;
            }`}},Xl=e=>`${e.format};${e.ceilMode};${e.autoPad};${e.kernelShape.length}`,N_=e=>`${Xl(e)};${e.countIncludePad}`,j_=e=>`${Xl(e)};${e.storageOrder};${e.dilations}`,Jl=e=>({format:e.format,autoPad:["NOTSET","VALID","SAME_UPPER","SAME_LOWER"][e.auto_pad],ceilMode:e.ceil_mode,kernelShape:e.kernel_shape,strides:e.strides,pads:e.pads}),Yl=(e,r,t,s)=>{let[o,n]=Kl(r,s,t),a=Oe("x",r.dataType,r.dims.length),i=a.type.value,l="value += x_val;",c="";o.countIncludePad?c+=`value /= ${i}(uniforms.kernelSize);`:c+=`value /= ${i}(i32(uniforms.kernelSize) - pad);`;let[p,d,u,_,f]=ql(n,o);p.push(...lt(r.dims,n));let b=["rank"];return{name:e,shaderCache:{hint:`${s.cacheKey};${u};${_};${f}`,inputDependencies:b},getRunData:()=>({outputs:[{dims:n,dataType:r.dataType}],dispatchGroup:{x:Math.ceil(Pe.size(n)/64)},programUniforms:p}),getShaderSource:A=>Ql(A,a,r.dims.length,n.length,o,l,c,0,d,u,_,f)}},V_=e=>{let r=e.count_include_pad!==0,t=Jl(e);if(t.ceilMode!==0)throw new Error("using ceil() in shape computation is not yet supported for AveragePool");let s={countIncludePad:r,...t,cacheKey:""};return{...s,cacheKey:N_(s)}},U_=(e,r)=>{To(e.inputs),e.compute(Yl("AveragePool",e.inputs[0],!1,r))},Zl={autoPad:"",ceilMode:0,countIncludePad:!1,kernelShape:[],strides:[],pads:[],storageOrder:0,dilations:[]},W_=e=>{let r=e.format;return{format:r,...Zl,cacheKey:r}},G_=(e,r)=>{To(e.inputs),e.compute(Yl("GlobalAveragePool",e.inputs[0],!0,r))},eu=(e,r,t,s)=>{let[o,n]=Kl(r,s,t),a=`
      value = max(x_val, value);
    `,i="",l=Oe("x",r.dataType,r.dims.length),c=["rank"],[p,d,u,_,f]=ql(n,o);return p.push(...lt(r.dims,n)),{name:e,shaderCache:{hint:`${s.cacheKey};${u};${_};${f}`,inputDependencies:c},getRunData:()=>({outputs:[{dims:n,dataType:r.dataType}],dispatchGroup:{x:Math.ceil(Pe.size(n)/64)},programUniforms:p}),getShaderSource:b=>Ql(b,l,r.dims.length,n.length,o,a,i,r.dataType===10?-65504:-1e5,d,u,_,f)}},H_=(e,r)=>{To(e.inputs),e.compute(eu("MaxPool",e.inputs[0],!1,r))},K_=e=>{let r=e.storage_order,t=e.dilations,s=Jl(e);if(r!==0)throw new Error("column major storage order is not yet supported for MaxPool");if(s.ceilMode!==0)throw new Error("using ceil() in shape computation is not yet supported for MaxPool");let o={storageOrder:r,dilations:t,...s,cacheKey:""};return{...o,cacheKey:j_(o)}},q_=e=>{let r=e.format;return{format:r,...Zl,cacheKey:r}},Q_=(e,r)=>{To(e.inputs),e.compute(eu("GlobalMaxPool",e.inputs[0],!0,r))}}),X_,J_,Y_,Z_,PT=Ue(()=>{Mt(),Pt(),ur(),Ct(),X_=(e,r)=>{if(e.length<2||e.length>3)throw new Error("DequantizeLinear requires 2 or 3 inputs.");if(e.length===3&&e[1].dims===e[2].dims)throw new Error("x-scale and x-zero-point must have the same shape.");if(e.length===3&&e[0].dataType!==e[2].dataType)throw new Error("x and x-zero-point must have the same data type.");if(e[0].dataType===6&&e.length>2)throw new Error("In the case of dequantizing int32 there is no zero point.");if(e[1].dims.length!==0&&e[1].dims.length!==1&&e[1].dims.length!==e[0].dims.length)throw new Error("scale input must be a scalar, a 1D tensor, or have the same rank as the input tensor.");if(e.length>2){if(e[0].dataType!==e[2].dataType)throw new Error("x and x-zero-point must have the same data type.");if(e[1].dims.length!==e[2].dims.length)throw new Error("scale and zero-point inputs must have the same rank.");if(!e[1].dims.map((t,s)=>t===e[2].dims[s]).reduce((t,s)=>t&&s,!0))throw new Error("scale and zero-point inputs must have the same shape.")}if(r.blockSize>0){if(e[1].dims.length===0||e[1].dims.length===1&&e[1].dims[0]===1)throw new Error("blockSize must be set only for block quantization.");if(!e[1].dims.map((o,n)=>n===r.axis||o===e[0].dims[n]).reduce((o,n)=>o&&n,!0))throw new Error("For block qunatization, scale input shape to match the input shape except for the axis");if(e[1].dims.length!==e[0].dims.length)throw new Error("For block qunatization the scale input rank must be the same as the x rank.");let t=e[0].dims[r.axis],s=e[1].dims[r.axis];if(r.blockSize<Math.ceil(t/s)||r.blockSize>Math.ceil(t/(s-1)-1))throw new Error("blockSize must be with in the range [ceil(dI / Si), ceil(dI / (Si - 1) - 1)].")}},J_=(e,r)=>{let t=Pe.normalizeAxis(r.axis,e[0].dims.length),s=e[0].dataType,o=s===3,n=e[0].dims,a=e[1].dataType,i=Pe.size(n),l=s===3||s===2,c=l?[Math.ceil(Pe.size(e[0].dims)/4)]:e[0].dims,p=e[1].dims,d=e.length>2?e[2]:void 0,u=d?l?[Math.ceil(Pe.size(d.dims)/4)]:d.dims:void 0,_=p.length===0||p.length===1&&p[0]===1,f=_===!1&&p.length===1,b=or(i),A=_&&(!l||b===4),g=A?b:1,y=A&&!l?b:1,C=Oe("input",l?12:s,c.length,y),x=Oe("scale",a,p.length),M=d?Oe("zero_point",l?12:s,u.length):void 0,T=at("output",a,n.length,g),v=[C,x];M&&v.push(M);let P=[c,p];d&&P.push(u);let F=[{type:12,data:i/g},{type:12,data:t},{type:12,data:r.blockSize},...lt(...P,n)],D=K=>{let U=[{name:"output_size",type:"u32"},{name:"axis",type:"u32"},{name:"block_size",type:"u32"}];return`
      ${K.registerUniforms(U).declareVariables(...v,T)}
      ${K.mainStart()}
          ${K.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_size")}
          let output_indices = ${T.offsetToIndices("global_idx")};

          // Set input x
          ${l?`
            let input = ${C.getByOffset("global_idx / 4")};
            let x_vec = ${o?"unpack4xI8(input)":"unpack4xU8(input)"};
            let x_value = ${g===1?"x_vec[global_idx % 4]":"x_vec"};`:`let x_value = ${C.getByOffset("global_idx")};`};

          // Set scale input
          ${_?`let scale_value= ${x.getByOffset("0")}`:f?`
            let scale_index = ${T.indicesGet("output_indices","uniforms.axis")};
            let scale_value= ${x.getByOffset("scale_index")};`:`
            var scale_indices: ${x.type.indices} = output_indices;
            let index = ${x.indicesGet("scale_indices","uniforms.axis")} / uniforms.block_size;
            ${x.indicesSet("scale_indices","uniforms.axis","index")};
            let scale_value= ${x.getByIndices("scale_indices")};`};

          // Set zero-point input
          ${M?_?l?`
                let zero_point_input = ${M.getByOffset("0")};
                let zero_point_vec =  ${o?"unpack4xI8(zero_point_input)":"unpack4xU8(zero_point_input)"};
                let zero_point_value= zero_point_vec[0]`:`let zero_point_value = ${M.getByOffset("0")}`:f?l?`
                let zero_point_index = ${T.indicesGet("output_indices","uniforms.axis")};
                let zero_point_input = ${M.getByOffset("zero_point_index / 4")};
                let zero_point_vec =  ${o?"unpack4xI8(zero_point_input)":"unpack4xU8(zero_point_input)"};
                let zero_point_value = zero_point_vec[zero_point_index % 4]`:`
                let zero_point_index = ${T.indicesGet("output_indices","uniforms.axis")};
                let zero_point_value = ${M.getByOffset("zero_point_index")};`:l?`
                let zero_point_offset = ${x.indicesToOffset("scale_indices")};
                let zero_point_input = ${M.getByOffset("zero_point_offset / 4")};
                let zero_point_vec = ${o?"unpack4xI8(zero_point_input)":"unpack4xU8(zero_point_input)"};
                let zero_point_value = zero_point_vec[zero_point_offset % 4];`:`let zero_point_value = ${M.getByIndices("scale_indices")};`:`let zero_point_value = ${l?o?"i32":"u32":C.type.value}(0);`};
      // Compute and write output
      ${T.setByOffset("global_idx",`${T.type.value}(x_value - zero_point_value) * scale_value`)};
      }`};return{name:"DequantizeLinear",shaderCache:{hint:r.cacheKey,inputDependencies:M?["rank","rank","rank"]:["rank","rank"]},getShaderSource:D,getRunData:()=>({outputs:[{dims:n,dataType:a}],dispatchGroup:{x:Math.ceil(i/g/64),y:1,z:1},programUniforms:F})}},Y_=(e,r)=>{X_(e.inputs,r),e.compute(J_(e.inputs,r))},Z_=e=>jt({axis:e.axis,blockSize:e.blockSize})}),eg,tg,rg,CT=Ue(()=>{bs(),Mt(),Ct(),eg=(e,r,t)=>{let s=e===r,o=e<r&&t<0,n=e>r&&t>0;if(s||o||n)throw new Error("Range these inputs' contents are invalid.")},tg=(e,r,t,s)=>{let o=Math.abs(Math.ceil((r-e)/t)),n=[o],a=o,i=[{type:12,data:a},{type:s,data:e},{type:s,data:t},...lt(n)],l=c=>{let p=at("output",s,n.length),d=p.type.value,u=[{name:"outputSize",type:"u32"},{name:"start",type:d},{name:"delta",type:d}];return`
        ${c.registerUniforms(u).declareVariables(p)}
        ${c.mainStart()}
        ${c.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.outputSize")}
        output[global_idx] = uniforms.start + ${d}(global_idx) * uniforms.delta;
      }`};return{name:"Range",shaderCache:{hint:`${s}`},getShaderSource:l,getRunData:()=>({outputs:[{dims:n,dataType:s}],dispatchGroup:{x:Math.ceil(a/64)},programUniforms:i})}},rg=e=>{let r=0,t=0,s=0;e.inputs[0].dataType===6?(r=e.inputs[0].getInt32Array()[0],t=e.inputs[1].getInt32Array()[0],s=e.inputs[2].getInt32Array()[0]):e.inputs[0].dataType===1&&(r=e.inputs[0].getFloat32Array()[0],t=e.inputs[1].getFloat32Array()[0],s=e.inputs[2].getFloat32Array()[0]),Jt.webgpu.validateInputContent&&eg(r,t,s),e.compute(tg(r,t,s,e.inputs[0].dataType),{inputs:[]})}}),sg,tu,ru,ng,og,ag,ST=Ue(()=>{Mt(),Pt(),ur(),Ct(),sg=(e,r,t,s)=>{if(e!=="none"&&s!=="i32"&&s!=="u32"&&s!=="f32")throw new Error(`Input ${s} is not supported with reduction ${e}.`);let o=`{
                var oldValue = 0;
                loop {
                  let newValueF32 =`,n=`;
                  let newValue = bitcast<i32>(newValueF32);
                  let res = atomicCompareExchangeWeak(&${r}, oldValue, newValue);
                  if res.exchanged {
                    break;
                  }
                  oldValue = res.old_value;
                }
              }`;switch(e){case"none":return`${r}=${t};`;case"add":return s==="i32"||s==="u32"?`atomicAdd(&${r}, bitcast<${s}>(${t}));`:`
              ${o}bitcast<${s}>(oldValue) + (${t})${n}`;case"max":return s==="i32"||s==="u32"?`atomicMax(&${r}, bitcast<${s}>(${t}));`:`
                ${o}max(bitcast<f32>(oldValue), (${t}))${n}`;case"min":return s==="i32"||s==="u32"?`atomicMin(&${r}, bitcast<${s}>(${t}));`:`${o}min(bitcast<${s}>(oldValue), (${t}))${n}`;case"mul":return`${o}(bitcast<${s}>(oldValue) * (${t}))${n}`;default:throw new Error(`Reduction ${e} is not supported.`)}},tu=(e,r)=>`${e===1?`
    let element_count_dim = uniforms.output_strides;
    let dim_value = uniforms.output_shape;`:`
    let element_count_dim = uniforms.output_strides[${r?"i - indices_start":"i"}];
    let dim_value = uniforms.output_shape[${r?"i - indices_start":"i"} + uniforms.last_index_dimension];`}
    
    if (index >= 0) {
      if (index >= i32(dim_value)) {
        index = i32(dim_value - 1);
      }
    } else {
      if (index < -i32(dim_value)) {
        index = 0;
      } else {
        index += i32(dim_value);
      }
    }
    data_offset += u32((u32(index) * element_count_dim));`,ru=(e,r,t)=>`for (var i = 0u; i < uniforms.num_updates_elements; i++) {
        let value = updates[uniforms.num_updates_elements * ${t?"global_idx":"idx"} + i];
        ${sg(e.reduction,"output[data_offset + i]","value",r)}
      }`,ng=(e,r)=>{let t=e[0].dims,s=e[1].dims,o=t,n=1,a=Math.ceil(Pe.size(s)/n),i=s[s.length-1],l=Pe.sizeFromDimension(t,i),c=Pe.sizeFromDimension(s,0)/i,p=[{type:12,data:a},{type:12,data:i},{type:12,data:l},...lt(e[1].dims,e[2].dims,o)],d=u=>{let _=Oe("indices",e[1].dataType,e[1].dims.length),f=Oe("updates",e[2].dataType,e[2].dims.length,n),b=r.reduction!=="none"&&r.reduction!==""?dp("output",e[0].dataType,o.length):at("output",e[0].dataType,o.length,n);return`
      ${u.registerUniform("output_size","u32").registerUniform("last_index_dimension","u32").registerUniform("num_updates_elements","u32").declareVariables(_,f,b)}
      ${u.mainStart()}
        ${u.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_size")}
  var hasDuplicates = false;
  if (${r.reduction==="none"}) {
    for (var i = 0; i < ${c}; i = i + 1) {
      for (var j = i + 1; j < ${c}; j = j + 1) {
        var index_i = i32(indices[i].x);
        var index_j = i32(indices[j].x);
        if (index_i == index_j) {
          hasDuplicates = true;
          break;
        }
      }
      if (hasDuplicates) {
        break;
      }
    }
  }

  if (${r.reduction==="none"} && hasDuplicates) {
    if (global_idx != 0u) {
      return;
    }
    // Process each index-update pair individually when duplicates exist
    for (var idx = 0u; idx < ${c}u; idx++) {
      var data_offset = 0u;
      for (var i = 0u; i < uniforms.last_index_dimension; i++) {
        var index = i32(indices[idx * uniforms.last_index_dimension + i].x);
        ${tu(t.length,!1)}
      }
      ${ru(r,b.type.value,!1)}
    }
    return;
  }

  var data_offset = 0u;
  var indices_start = uniforms.last_index_dimension * global_idx;
  var indices_end = indices_start + uniforms.last_index_dimension;
  for (var i = indices_start; i < indices_end; i++) {
    var index = i32(indices[i].x);
    ${tu(t.length,!0)}
  }
  ${ru(r,b.type.value,!0)}
  }`};return{name:"ScatterND",shaderCache:{hint:`${r.cacheKey}_${r.reduction}`,inputDependencies:["rank","rank"]},getRunData:()=>({outputs:[{dims:o,dataType:e[0].dataType}],dispatchGroup:{x:Math.ceil(a/64)},programUniforms:p}),getShaderSource:d}},og=e=>jt({reduction:e.reduction}),ag=(e,r)=>{e.compute(ng(e.inputs,r),{inputs:[e.inputs[1],e.inputs[2]],outputs:[]})}}),ig,lg,ug,su,cg,dg,pg,hg,mg,fg,_g,gg,nu,wg,Mg,bg,yg,vg,xg,Tg,$T=Ue(()=>{Mt(),Pt(),ur(),Ct(),ig=(e,r)=>{if(e.every(t=>t>0||(()=>{throw new Error("Resize requires scales input values to be positive")})),e.length>0){if(r.mode==="linear"){if(!(e.length===2||e.length===3||e.length===4&&e[0]===1&&e[1]===1||e.length===4&&e[0]===1&&e[3]===1||e.length===5&&e[0]===1&&e[1]===1))throw new Error(`For linear mode, Resize requires scales to be 2D, 3D, 4D with either two outermost or one innermost and
            one outermost scale values equal to 1, or 5D with two outermost scale values equal to 1`)}else if(r.mode==="cubic"&&!(e.length===2||e.length===4&&e[0]===1&&e[1]===1||e.length===4&&e[0]===1&&e[3]===1))throw new Error("Resize requires scales input size to be 2 or 4 for cubic mode")}},lg=(e,r,t)=>{r.every(o=>o>=0&&o<t||(()=>{throw new Error("Resize requires axes input values to be positive and less than rank")}));let s=new Array(t).fill(1);return r.forEach((o,n)=>s[o]=e[n]),s},ug=(e,r,t,s,o,n)=>{let[a,i,l]=t>10?[1,2,3]:[-1,e.length>1?1:-1,-1],c=e[0].dims.length;if(a>0&&e.length>a&&e[a].dims.length>0)e[a].getFloat32Array().forEach(p=>n.push(p));else if(r.coordinateTransformMode==="tf_crop_and_resize")throw new Error("Resize requires RoI input to be specified when coordinateTransformMode is tfCropAndResize");if(i>0&&e.length>i&&e[i].dims.length===1&&e[i].dims[0]>0){if(e[i].getFloat32Array().forEach(p=>s.push(p)),s.length!==0&&s.length!==c&&t>=18&&s.length!==r.axes.length)throw new Error("Resize requires scales input size to be same as input rank or axes size for opset 18 and up");ig(s,r),r.axes.length>0&&lg(s,r.axes,c).forEach((p,d)=>s[d]=p)}if(l>0&&e.length>l&&e[l].dims.length===1&&e[l].dims[0]>0&&(e[l].getBigInt64Array().forEach(p=>o.push(Number(p))),o.length!==0&&o.length!==c&&t>=18&&o.length!==r.axes.length))throw new Error("Resize requires sizes input size to be same as input rank or axes size for opset 18 and up");if(r.axes.length>0){if(s.length!==0&&s.length!==r.axes.length)throw new Error('Resize requires "scales" input size to be of axes rank when axes attributes is specified');if(o.length!==0&&o.length!==r.axes.length)throw new Error('Resize requires "sizes" input size to be of rank axes rank when axes attributes is specified')}if(typeof s<"u"&&typeof o<"u"&&s.length>0&&o.length>c)throw new Error("Resize requires only of scales or sizes to be specified")},su=(e,r,t,s)=>`
  // The whole part and the fractional part are calculated separately due to inaccuracy of floating
  // point division. As an example, f32(21) / f32(7) may evaluate to 2.99... instead of 3, causing an
  // offset-by-one error later in floor().
  let big = (${e}) * (${r});
  let whole = ${s}(big / (${t}));
  let fract = ${s}(big % (${t})) / ${s}(${t});
  return whole + fract;
`,cg=(e,r)=>`fn getOriginalCoordinateFromResizedCoordinate(xResized: u32, xScale: f32, lengthResized: u32,
     lengthOriginal: u32, roiStart: f32, roiEnd: f32) -> ${r} { `+(()=>{switch(e){case"asymmetric":return`
          if (xScale < 1.0 || floor(xScale) != xScale) {
            return ${r}(xResized) / ${r}(xScale);
          } else {
            ${su("xResized","lengthOriginal","lengthResized",r)}
          }
        `;case"pytorch_half_pixel":return`if (lengthResized > 1) {
                    return (${r}(xResized) + 0.5) / ${r}(xScale) - 0.5;
                  } else {
                    return 0.0;
                  }`;case"tf_half_pixel_for_nn":return`return (${r}(xResized) + 0.5) / ${r}(xScale);`;case"align_corners":return`if (lengthResized == 1) {
                    return 0.0;
                  } else {
                    ${su("xResized","lengthOriginal - 1","lengthResized - 1",r)}
                  }`;case"tf_crop_and_resize":return`if (lengthResized > 1) {
                    return ${r}(roiStart) * ${r}(lengthOriginal - 1) +
                        (${r}(xResized) * ${r}(roiEnd - roiStart) * ${r}(lengthOriginal - 1)) /
                        ${r}(lengthResized - 1);
                  } else {
                    return 0.5 * ${r}(roiStart + roiEnd) * ${r}(lengthOriginal - 1);
                  }`;case"half_pixel_symmetric":return`const outputWidth = ${r}xScale * ${r}(lengthResized);
                  const adjustment = ${r}(lengthResized) / outputWidth;
                  const center = ${r}(lengthOriginal) / 2;
                  const offset = center * (1 - adjustment);
                  return offset + ((${r}(xResized) + 0.5) / ${r}(xScale)) - 0.5;`;case"half_pixel":return`return ((${r}(xResized) + 0.5) / ${r}(xScale)) - 0.5;`;default:throw new Error(`Coordinate transform mode ${e} is not supported`)}})()+"}",dg=(e,r,t)=>`fn getNearestPixelFromOriginal(xOriginal: ${t}, isDownSample: bool) -> ${t} {`+(()=>{switch(e){case"round_prefer_ceil":return"if (fract(xOriginal) == 0.5) {             return ceil(xOriginal);           } else {             return round(xOriginal);           }";case"floor":return"return floor(xOriginal);";case"ceil":return"return ceil(xOriginal);";case"round_prefer_floor":return"if (fract(xOriginal) == 0.5) {                     return floor(xOriginal);                   } else {                     return round(xOriginal);                   }";case"simple":default:if(r<11)return"if (isDownSample)                     {                       return ceil(xOriginal);                     } else {                       return xOriginal;                     }";throw new Error(`Nearest mode ${e} is not supported`)}})()+"}",pg=(e,r,t)=>{let s=new Array(t).fill(0).concat(new Array(t).fill(1)),o=e.length===0?s:e.slice();return r.length>0?(r.forEach((n,a)=>{s[n]=o[a],s[a+t]=o[r.length+a]}),s):o},hg=(e,r,t,s)=>{let o=[];if(t.length>0)if(s.length>0){if(e.forEach(n=>o.push(n)),Math.max(...s)>e.length)throw new Error("axes is out of bound");s.forEach((n,a)=>o[n]=t[a])}else t.forEach(n=>o.push(n));else{if(r.length===0)throw new Error("Resize requires either scales or sizes.");o=e.map((n,a)=>Math.round(n*r[a]))}return o},mg=(e,r,t)=>{let s=(()=>{switch(t.keepAspectRatioPolicy){case"not_larger":return t.axes.length>0?Math.min(...t.axes.map(n=>r[n]),Number.MAX_VALUE):Math.min(...r,Number.MAX_VALUE);case"not_smaller":return t.axes.length>0?Math.max(...t.axes.map(n=>r[n]),Number.MIN_VALUE):Math.max(...r,Number.MIN_VALUE);default:throw new Error(`Keep aspect ratio policy ${t.keepAspectRatioPolicy} is not supported`)}})();r.fill(1,0,r.length);let o=e.slice();return t.axes.length>0?(t.axes.forEach(n=>r[n]=s),t.axes.forEach(n=>o[n]=Math.round(e[n]*r[n]))):(r.fill(s,0,r.length),o.forEach((n,a)=>o[a]=Math.round(n*r[a]))),o},fg=(e,r,t,s,o)=>`
    fn calculateOriginalIndicesFromOutputIndices(output_indices: ${e.type.indices}) -> array<${e.type.value}, ${t.length}> {
      var original_indices: array<${e.type.value}, ${t.length}>;
      for (var i:u32 = 0; i < ${t.length}; i++) {
        var output_index = ${e.indicesGet("output_indices","i")};
        var scale = ${it("uniforms.scales","i",s)};
        var roi_low = ${it("uniforms.roi","i",o)};
        var roi_hi = ${it("uniforms.roi",`i + ${r.length}`,o)};
        if (scale == 1.0) {
          original_indices[i] = ${e.type.value}(output_index);
        } else {
          var input_shape_i = ${it("uniforms.input_shape","i",r.length)};
          var output_shape_i = ${it("uniforms.output_shape","i",t.length)};
          original_indices[i] = getOriginalCoordinateFromResizedCoordinate(output_index, scale, output_shape_i,
                                                                           input_shape_i, roi_low, roi_hi);
        }
      }
      return original_indices;
    }`,_g=(e,r,t,s,o,n,a)=>`
    fn calculateInputIndicesFromOutputIndices(output_indices: ${r.type.indices}) -> ${e.type.indices} {
      var input_indices: ${e.type.indices};
      for (var i:u32 = 0; i < ${s.length}; i++) {
        var output_index = ${r.indicesGet("output_indices","i")};
        var input_index: u32;
        var scale = ${it("uniforms.scales","i",o)};
        if (scale == 1.0) {
          input_index = output_index;
        } else {
          var roi_low = ${it("uniforms.roi","i",n)};
          var roi_hi = ${it("uniforms.roi",`i + ${t.length}`,n)};
          var input_shape_i = ${it("uniforms.input_shape","i",t.length)};
          var output_shape_i = ${it("uniforms.output_shape","i",s.length)};
          var original_idx = getOriginalCoordinateFromResizedCoordinate(output_index, scale, output_shape_i,
                                                                        input_shape_i, roi_low, roi_hi);
          if (!${a} || (original_idx >= 0 && original_idx < ${r.type.value}(input_shape_i))) {
            if (original_idx < 0) {
              input_index = 0;
            } else if (original_idx > ${r.type.value}(input_shape_i - 1)) {
              input_index = input_shape_i - 1;
            } else {
              input_index = u32(getNearestPixelFromOriginal(original_idx, scale < 1));
            }
          } else {
            input_index = u32(original_idx);
          }
        }
        ${e.indicesSet("input_indices","i","input_index")}
      }
      return input_indices;
    }`,gg=(e,r)=>`
    fn checkInputIndices(input_indices: ${e.type.indices}) -> bool {
      for (var i:u32 = 0; i < ${r.length}; i++) {
        var input_index = ${e.indicesGet("input_indices","i")};
        if (input_index < 0 || input_index >= ${it("uniforms.input_shape","i",r.length)}) {
          return false;
        }
      }
      return true;
    }`,nu=(e,r,t,s)=>e.rank>s?`
    ${e.indicesSet("input_indices",r,"channel")};
    ${e.indicesSet("input_indices",t,"batch")};
`:"",wg=(e,r,t,s,o)=>{let[n,a,i,l]=t.length===2?[-1,0,1,-1]:[0,2,3,1],c=e.type.value;return`
    fn getInputValue(batch: u32, channel: u32, row: u32, col: u32) -> ${c} {
      var input_indices: ${e.type.indices};
      ${e.indicesSet("input_indices",a,`max(0, min(row, ${t[a]} - 1))`)};
      ${e.indicesSet("input_indices",i,`max(0, min(col, ${t[i]} - 1))`)};
      ${nu(e,l,n,2)}
      return ${e.getByIndices("input_indices")};
    }

    fn bilinearInterpolation(output_indices: ${r.type.indices}) -> ${c} {
      var originalIndices = calculateOriginalIndicesFromOutputIndices(output_indices);
      var row:${c} = originalIndices[${a}];
      var col:${c} = originalIndices[${i}];
      ${s?`if (row < 0 || row > (${t[a]} - 1) || col < 0 || col > (${t[i]} - 1)) {
        return ${o};
      }`:""};
      row = max(0, min(row, ${t[a]} - 1));
      col = max(0, min(col, ${t[i]} - 1));
      var row1: u32 = u32(row);
      var col1: u32 = u32(col);
      var row2: u32 = u32(row + 1);
      var col2: u32 = u32(col + 1);
      var channel: u32 = ${t.length>2?`u32(originalIndices[${l}])`:"0"};
      var batch: u32 =  ${t.length>2?`u32(originalIndices[${n}])`:"0"};
      var x11: ${c} = getInputValue(batch, channel, row1, col1);
      var x12: ${c} = getInputValue(batch, channel, row1, col2);
      var x21: ${c} = getInputValue(batch, channel, row2, col1);
      var x22: ${c} = getInputValue(batch, channel, row2, col2);
      var dx1: ${c} = abs(row - ${c}(row1));
      var dx2: ${c} = abs(${c}(row2) - row);
      var dy1: ${c} = abs(col - ${c}(col1));
      var dy2: ${c} = abs(${c}(col2) - col);
      if (row1 == row2) {
        dx1 = 0.5;
        dx2 = 0.5;
      }
      if (col1 == col2) {
        dy1 = 0.5;
        dy2 = 0.5;
      }
      return (x11 * dx2 * dy2 + x12 * dx2 * dy1 + x21 * dx1 * dy2 + x22 * dx1 * dy1);
    }`},Mg=(e,r,t,s,o,n,a,i,l,c)=>{let p=t.length===2,[d,u]=p?[0,1]:[2,3],_=e.type.value,f=b=>{let A=b===d?"row":"col";return`
      fn ${A}CubicInterpolation(input_indices: ${e.type.indices}, output_indices: ${r.type.indices}) -> ${_} {
        var output_index = ${r.indicesGet("output_indices",b)};
        var originalIdx: ${_} = getOriginalCoordinateFromResizedCoordinate(output_index, ${o[b]},
        ${s[b]}, ${t[b]}, ${n[b]}, ${n[b]} + ${t.length});
        var fractOriginalIdx: ${_} = originalIdx - floor(originalIdx);
        var coefs = getCubicInterpolationCoefs(fractOriginalIdx);

        if (${i} && (originalIdx < 0 || originalIdx > (${t[b]} - 1))) {
          return ${l};
        }
        var data: array<${_}, 4> = array<${_}, 4>(0.0, 0.0, 0.0, 0.0);
        for (var i: i32 = -1; i < 3; i++) {
          var ${A}: ${_} = originalIdx + ${_}(i);
          if (${A} < 0 || ${A} >= ${t[b]}) {
            ${c?`coefs[i + 1] = 0.0;
                        continue;`:i?`return ${l};`:`${A} = max(0, min(${A}, ${t[b]} - 1));`};
          }
        var input_indices_copy: ${e.type.indices} = input_indices;
          ${e.indicesSet("input_indices_copy",b,`u32(${A})`)};
          data[i + 1] = ${b===d?e.getByIndices("input_indices_copy"):"rowCubicInterpolation(input_indices_copy, output_indices)"};
        }
        return cubicInterpolation1D(data, coefs);
      }`};return`
    ${f(d)};
    ${f(u)};
  fn getCubicInterpolationCoefs(s: ${_}) -> array<${_}, 4> {
    var absS = abs(s);
    var coeffs: array<${_}, 4> = array<${_}, 4>(0.0, 0.0, 0.0, 0.0);
    var oneMinusAbsS: ${_} = 1.0 - absS;
    var twoMinusAbsS: ${_} = 2.0 - absS;
    var onePlusAbsS: ${_} = 1.0 + absS;
    coeffs[0] = ((${a} * onePlusAbsS - 5 * ${a}) * onePlusAbsS + 8 * ${a}) * onePlusAbsS - 4 * ${a};
    coeffs[1] = ((${a} + 2) * absS - (${a} + 3)) * absS * absS + 1;
    coeffs[2] = ((${a} + 2) * oneMinusAbsS - (${a} + 3)) * oneMinusAbsS * oneMinusAbsS + 1;
    coeffs[3] = ((${a} * twoMinusAbsS - 5 * ${a}) * twoMinusAbsS + 8 * ${a}) * twoMinusAbsS - 4 * ${a};
    return coeffs;
  }

  fn cubicInterpolation1D(x: array<${_}, 4>, coefs: array<${_}, 4>) -> ${_} {
    var coefsSum: ${_} = coefs[0] + coefs[1] + coefs[2] + coefs[3];
    return (x[0] * coefs[0] + x[1] * coefs[1]+ x[2] * coefs[2]+ x[3] * coefs[3]) / coefsSum;
  }

  fn bicubicInterpolation(output_indices: ${r.type.indices}) -> ${_} {
    var input_indices: ${e.type.indices} = output_indices;
    return colCubicInterpolation(input_indices, output_indices);
  }
    `},bg=(e,r,t,s,o)=>{let[n,a,i,l,c]=t.length===3?[-1,0,1,2,-1]:[0,2,3,4,1],p=e.type.value;return`
    fn getInputValue(batch: u32, channel: u32, depth:u32, height: u32, width: u32) -> ${p} {
      var input_indices: ${e.type.indices};
      ${e.indicesSet("input_indices",a,`max(0, min(depth, ${t[a]} - 1))`)};
      ${e.indicesSet("input_indices",i,`max(0, min(height, ${t[i]} - 1))`)};
      ${e.indicesSet("input_indices",l,`max(0, min(width, ${t[l]} - 1))`)};
      ${nu(e,c,n,3)}
      return ${e.getByIndices("input_indices")};
    }

    fn trilinearInterpolation(output_indices: ${r.type.indices}) -> ${p} {
      var originalIndices = calculateOriginalIndicesFromOutputIndices(output_indices);
      var depth:${p} = originalIndices[${a}];
      var height:${p} = originalIndices[${i}];
      var width:${p} = originalIndices[${l}];
      ${s?`if (depth < 0 || depth > (${t[a]} - 1) || height < 0 || height > (${t[i]} - 1) || width < 0 || (width > ${t[l]} - 1)) {
      return ${o};
        }`:""};

    depth = max(0, min(depth, ${t[a]} - 1));
      height = max(0, min(height, ${t[i]} - 1));
      width = max(0, min(width, ${t[l]} - 1));
      var depth1: u32 = u32(depth);
      var height1: u32 = u32(height);
      var width1: u32 = u32(width);
      var depth2: u32 = u32(depth + 1);
      var height2: u32 = u32(height + 1);
      var width2: u32 = u32(width + 1);
      var channel: u32 = ${t.length>3?`u32(originalIndices[${c}])`:"0"};
      var batch: u32 =  ${t.length>3?`u32(originalIndices[${n}])`:"0"};

      var x111: ${p} = getInputValue(batch, channel, depth1, height1, width1);
      var x112: ${p} = getInputValue(batch, channel, depth1, height1, width2);
      var x121: ${p} = getInputValue(batch, channel, depth1, height2, width1);
      var x122: ${p} = getInputValue(batch, channel, depth1, height2, width2);
      var x211: ${p} = getInputValue(batch, channel, depth2, height1, width1);
      var x212: ${p} = getInputValue(batch, channel, depth2, height1, width2);
      var x221: ${p} = getInputValue(batch, channel, depth2, height2, width1);
      var x222: ${p} = getInputValue(batch, channel, depth2, height2, width2);
      var dx1: ${p} = abs(depth - ${p}(depth1));
      var dx2: ${p} = abs(${p}(depth2) - depth);
      var dy1: ${p} = abs(height - ${p}(height1));
      var dy2: ${p} = abs(${p}(height2) - height);
      var dz1: ${p} = abs(width - ${p}(width1));
      var dz2: ${p} = abs(${p}(width2) - width);
      if (depth1 == depth2) {
        dx1 = 0.5;
        dx2 = 0.5;
      }
      if (height1 == height2) {
        dy1 = 0.5;
        dy2 = 0.5;
      }
      if (width1 == width2) {
        dz1 = 0.5;
        dz2 = 0.5;
      }
      return (x111 * dx2 * dy2 * dz2 + x112 * dx2 * dy2 * dz1 + x121 * dx2 * dy1 *dz2 + x122 * dx2 * dy1 * dz1 +
              x211 * dx1 * dy2 * dz2 + x212 * dx1 * dy2 * dz1 + x221 * dx1 * dy1 *dz2 + x222 * dx1 * dy1 * dz1);
    }`},yg=(e,r,t,s,o,n)=>{let a=e.dims,i=pg(n,r.axes,a.length),l=hg(a,s,o,r.axes),c=s.slice();s.length===0&&(c=a.map((y,C)=>y===0?1:l[C]/y),r.keepAspectRatioPolicy!=="stretch"&&(l=mg(a,c,r)));let p=at("output",e.dataType,l.length),d=Oe("input",e.dataType,a.length),u=Pe.size(l),_=a.length===l.length&&a.every((y,C)=>y===l[C]),f=r.coordinateTransformMode==="tf_crop_and_resize",b=r.extrapolationValue,A=d.type.value,g=y=>`
      ${_?"":`
      ${cg(r.coordinateTransformMode,A)};
      ${(()=>{switch(r.mode){case"nearest":return`
              ${gg(d,a)};
              ${dg(r.nearestMode,t,A)};
              ${_g(d,p,a,l,c.length,i.length,f)};
              `;case"linear":return`
              ${fg(p,a,l,c.length,i.length)};
              ${(()=>{if(a.length===2||a.length===4)return`${wg(d,p,a,f,b)}`;if(a.length===3||a.length===5)return`${bg(d,p,a,f,b)}`;throw Error("Linear mode only supports input dims 2, 3, 4 and 5 are supported in linear mode.")})()};
            `;case"cubic":return`
            ${(()=>{if(a.length===2||a.length===4)return`${Mg(d,p,a,l,c,i,r.cubicCoeffA,f,r.extrapolationValue,r.excludeOutside)}`;throw Error("Cubic mode only supports input dims 2 and 4 are supported in linear mode.")})()};
            `;default:throw Error("Invalid resize mode")}})()};
      `}
      ${y.registerUniform("output_size","u32").registerUniform("scales","f32",c.length).registerUniform("roi","f32",i.length).declareVariables(d,p)}
      ${y.mainStart()}
        ${y.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_size")}
        ${_?"output[global_idx] = input[global_idx];":`
        let output_indices = ${p.offsetToIndices("global_idx")};
        var input_indices: ${d.type.indices};
        ${(()=>{switch(r.mode){case"nearest":return`input_indices = calculateInputIndicesFromOutputIndices(output_indices);
                if (checkInputIndices(input_indices)) {
                  output[global_idx] = ${d.getByIndices("input_indices")};
                } else {
                  output[global_idx] = ${r.extrapolationValue};
                }`;case"linear":return`output[global_idx] = ${a.length===2||a.length===4?"bilinearInterpolation":"trilinearInterpolation"}(output_indices);`;case"cubic":return"output[global_idx] = bicubicInterpolation(output_indices);";default:throw Error(`Unsupported resize mode: ${r.mode}`)}})()};
`}
      }`;return{name:"Resize",shaderCache:{hint:`${r.cacheKey}|${t}|${c.length>0?r.mode==="cubic"?c:c.length:""}|${o.length>0?o:""}|${i.length>0?i:""}|${_}|${r.mode==="nearest"?a.length:a}`,inputDependencies:["rank"]},getShaderSource:g,getRunData:()=>({outputs:[{dims:l,dataType:e.dataType}],dispatchGroup:{x:Math.ceil(u/64)},programUniforms:[{type:12,data:u},{type:1,data:c},{type:1,data:i},...lt(a,l)]})}},vg=e=>{let r=e.customDataBuffer;return new Uint32Array(r,r.byteOffset,1)[0]},xg=(e,r)=>{let t=[],s=[],o=[],n=vg(e);if(r.antialias!==0)throw Error("Only default value (0) for Antialias attribute is supported");ug(e.inputs,r,n,t,s,o),e.compute(yg(e.inputs[0],r,n,t,s,o),{inputs:[0]})},Tg=e=>{let r=e.antialias,t=e.axes,s=e.coordinateTransformMode,o=e.cubicCoeffA,n=e.excludeOutside!==0,a=e.extrapolationValue,i=e.keepAspectRatioPolicy,l=e.mode,c=e.nearestMode===""?"simple":e.nearestMode;return jt({antialias:r,axes:t,coordinateTransformMode:s,cubicCoeffA:o,excludeOutside:n,extrapolationValue:a,keepAspectRatioPolicy:i,mode:l,nearestMode:c})}}),Eg,Pg,Cg,AT=Ue(()=>{Mt(),Pt(),Ct(),Eg=e=>{if(!e||e.length<3)throw new Error("layerNorm requires at least 3 inputs.");let r=e[0],t=e[1],s=e[2];if(r.dataType!==t.dataType||r.dataType!==s.dataType)throw new Error("All inputs must have the same data type");if(r.dims.length!==3&&r.dims.length!==2)throw new Error("Input must be 2D or 3D");if(t.dims.length!==3&&t.dims.length!==2)throw new Error("Skip must be 2D or 3D");let o=r.dims[r.dims.length-1],n=r.dims[r.dims.length-2];if(t.dims[t.dims.length-1]!==o)throw new Error("Skip must have the same hidden size as input");if(t.dims[t.dims.length-2]!==n)throw new Error("Skip must have the same sequence length as input");if(s.dims.length!==1)throw new Error("Gamma must be 1D");if(s.dims[s.dims.length-1]!==o)throw new Error("Gamma must have the same hidden size as input");if(e.length>3){let a=e[3];if(a.dims.length!==1)throw new Error("Beta must be 1D");if(a.dims[a.dims.length-1]!==o)throw new Error("Beta must have the same hidden size as input")}if(e.length>4){let a=e[4];if(a.dims.length!==1)throw new Error("Bias must be 1D");if(a.dims[a.dims.length-1]!==o)throw new Error("Bias must have the same hidden size as input")}},Pg=(e,r,t,s)=>{let o=r.simplified,n=e[0].dims,a=Pe.size(n),i=n,l=a,c=n.slice(-1)[0],p=s?n.slice(0,-1).concat(1):[],d=!o&&e.length>3,u=e.length>4,_=s&&t>1,f=s&&t>2,b=t>3,A=64,g=or(c),y=[{type:12,data:l},{type:12,data:g},{type:12,data:c},{type:1,data:r.epsilon}],C=M=>{let T=[{name:"output_size",type:"u32"},{name:"components",type:"u32"},{name:"hidden_size",type:"u32"},{name:"epsilon",type:"f32"}],v=[Oe("x",e[0].dataType,e[0].dims,g),Oe("skip",e[1].dataType,e[1].dims,g),Oe("gamma",e[2].dataType,e[2].dims,g)];d&&v.push(Oe("beta",e[3].dataType,e[3].dims,g)),u&&v.push(Oe("bias",e[4].dataType,e[4].dims,g)),v.push(at("output",e[0].dataType,i,g)),_&&v.push(at("mean_output",1,p)),f&&v.push(at("inv_std_output",1,p)),b&&v.push(at("input_skip_bias_sum",e[0].dataType,i,g));let P=Sr(e[0].dataType),F=Sr(1,g);return`

      ${M.registerUniforms(T).declareVariables(...v)}
      var<workgroup> sum_shared : array<${F}, ${A}>;
      var<workgroup> sum_squared_shared : array<${F}, ${A}>;

      ${M.mainStart([A,1,1])}
        let ix = local_id.x;
        let iy = global_id.x / ${A};

        let hidden_size_vectorized: u32 = uniforms.hidden_size / uniforms.components;
        var stride = hidden_size_vectorized / ${A};
        let offset = ix * stride + iy * hidden_size_vectorized;
        let offset1d = stride * ix;
        if (ix == ${A-1}) {
          stride = hidden_size_vectorized - stride * ix;
        }
        for (var i: u32 = 0; i < stride; i++) {
          let skip_value = skip[offset + i];
          let bias_value = ${u?"bias[offset1d + i]":P+"(0.0)"};
          let input_value = x[offset + i];
          let value = input_value + skip_value + bias_value;
          ${b?"input_skip_bias_sum[offset + i] = value;":""}
          output[offset + i] = value;
          let f32_value = ${Kn(P,g,"value")};
          sum_shared[ix] += f32_value;
          sum_squared_shared[ix] += f32_value * f32_value;
        }
        workgroupBarrier();

        var reduce_size : u32 = ${A};
        for (var curr_size = reduce_size >> 1;  curr_size > 0; curr_size = reduce_size >> 1) {
          reduce_size = curr_size + (reduce_size & 1);
          if (ix < curr_size) {
            sum_shared[ix] += sum_shared[ix + reduce_size];
            sum_squared_shared[ix] += sum_squared_shared[ix + reduce_size];
          }
          workgroupBarrier();
        }

        let sum = sum_shared[0];
        let square_sum = sum_squared_shared[0];
        let mean = ${Js("sum",g)} / f32(uniforms.hidden_size);
        let inv_std_dev = inverseSqrt(${Js("square_sum",g)} / f32(uniforms.hidden_size) ${o?"":"- mean * mean"} + uniforms.epsilon);
        ${_?"mean_output[global_idx] = mean;":""}
        ${f?"inv_std_output[global_idx] = inv_std_dev;":""}

        for (var i: u32 = 0; i < stride; i++) {
          output[offset + i] = (output[offset + i] ${o?"":`- ${P}(mean)`}) *
            ${P}(inv_std_dev) * gamma[offset1d + i]
            ${d?"+ beta[offset1d + i]":""};
        }
      }`},x=[{dims:i,dataType:e[0].dataType}];return t>1&&x.push({dims:p,dataType:1}),t>2&&x.push({dims:p,dataType:1}),t>3&&x.push({dims:n,dataType:e[0].dataType}),{name:"SkipLayerNormalization",shaderCache:{hint:`${g};${_};${f};${b}`,inputDependencies:e.map((M,T)=>"type")},getShaderSource:C,getRunData:()=>({outputs:x,dispatchGroup:{x:Math.ceil(l/c)},programUniforms:y})}},Cg=(e,r)=>{Eg(e.inputs);let t=[0];e.outputCount>1&&t.push(-3),e.outputCount>2&&t.push(-3),e.outputCount>3&&t.push(3),e.compute(Pg(e.inputs,r,e.outputCount,!1),{outputs:t})}}),Sg,Eo,$g,ou,Ag,kg,Ig,Fg,kT=Ue(()=>{Mt(),Pt(),ur(),Ct(),Sg=(e,r)=>{if(!e||e.length<1)throw new Error("too few inputs");if(r.axes.length!==0){if(r.axes.length!==r.starts.length||r.axes.length!==r.ends.length)throw new Error("axes, starts and ends must have the same length")}else if(r.starts.length!==r.ends.length)throw new Error("starts and ends must have the same length");e.slice(1).forEach((t,s)=>{if(e[s+1].dataType!==6&&e[s+1].dataType!==7)throw new Error(`Input ${s} must be an array of int32 or int64`)})},Eo=(e,r)=>{let t=[];if(e.length>r)if(e[r].dataType===7)e[r].getBigInt64Array().forEach(s=>t.push(Number(s)));else if(e[r].dataType===6)e[r].getInt32Array().forEach(s=>t.push(Number(s)));else throw new Error(`Input ${r} must be an array of int32 or int64`);return t},$g=(e,r)=>{if(e.length>1){let t=Eo(e,1),s=Eo(e,2),o=Eo(e,3);return o.length===0&&(o=[...Array(e[0].dims.length).keys()]),jt({starts:t,ends:s,axes:o})}else return r},ou=(e,r,t,s,o)=>{let n=e;return e<0&&(n+=t[s[r]]),o[r]<0?Math.max(0,Math.min(n,t[s[r]]-1)):Math.max(0,Math.min(n,t[s[r]]))},Ag=(e,r,t)=>`fn calculateInputIndices(output_indices: ${r.type.indices}) -> ${e.type.indices} {
          var input_indices: ${e.type.indices};
          var carry = 0u;
          for (var i = ${t.length}; i >= 0; i--) {
            let input_shape_i = ${it("uniforms.input_shape","i",t.length)};
            let steps_i = ${it("uniforms.steps","i",t.length)};
            let signs_i = ${it("uniforms.signs","i",t.length)};
            let starts_i = ${it("uniforms.starts","i",t.length)};
            var output_index = ${r.indicesGet("output_indices","i")};
            var input_index = output_index * steps_i + starts_i + carry;
            carry = input_index / input_shape_i;
            input_index = input_index % input_shape_i;
            if (signs_i < 0) {
              input_index = input_shape_i - input_index - 1u + starts_i;
            }
            ${e.indicesSet("input_indices","i","input_index")};
          }
          return input_indices;
      }`,kg=(e,r)=>{let t=e[0].dims,s=Pe.size(t),o=r.axes.length>0?Pe.normalizeAxes(r.axes,t.length):[...Array(t.length).keys()],n=Eo(e,4);n.forEach(g=>g!==0||(()=>{throw new Error("step cannot be 0")})),n.length===0&&(n=Array(o.length).fill(1));let a=r.starts.map((g,y)=>ou(g,y,t,o,n)),i=r.ends.map((g,y)=>ou(g,y,t,o,n));if(o.length!==a.length||o.length!==i.length)throw new Error("start, ends and axes should have the same number of elements");if(o.length!==t.length)for(let g=0;g<t.length;++g)o.includes(g)||(a.splice(g,0,0),i.splice(g,0,t[g]),n.splice(g,0,1));let l=n.map(g=>Math.sign(g));n.forEach((g,y,C)=>{if(g<0){let x=(i[y]-a[y])/g,M=a[y],T=M+x*n[y];a[y]=T,i[y]=M,C[y]=-g}});let c=t.slice(0);o.forEach((g,y)=>{c[g]=Math.ceil((i[g]-a[g])/n[g])});let p={dims:c,dataType:e[0].dataType},d=at("output",e[0].dataType,c.length),u=Oe("input",e[0].dataType,e[0].dims.length),_=Pe.size(c),f=[{name:"outputSize",type:"u32"},{name:"starts",type:"u32",length:a.length},{name:"signs",type:"i32",length:l.length},{name:"steps",type:"u32",length:n.length}],b=[{type:12,data:_},{type:12,data:a},{type:6,data:l},{type:12,data:n},...lt(e[0].dims,c)],A=g=>`
      ${g.registerUniforms(f).declareVariables(u,d)}
        ${Ag(u,d,t)}
        ${g.mainStart()}
          ${g.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.outputSize")}
          let output_indices = ${d.offsetToIndices("global_idx")};
          let input_indices = calculateInputIndices(output_indices);
          ${d.setByOffset("global_idx",u.getByIndices("input_indices"))}
      }`;return{name:"Slice",shaderCache:{hint:`${l.length}_${a.length}_${n.length}`,inputDependencies:["rank"]},getShaderSource:A,getRunData:()=>({outputs:[p],dispatchGroup:{x:Math.ceil(s/64)},programUniforms:b})}},Ig=(e,r)=>{Sg(e.inputs,r);let t=$g(e.inputs,r);e.compute(kg(e.inputs,t),{inputs:[0]})},Fg=e=>{let r=e.starts,t=e.ends,s=e.axes;return jt({starts:r,ends:t,axes:s})}}),Og,Dg,Lg,zg,IT=Ue(()=>{Mt(),Pt(),ur(),Ys(),Ct(),Og=e=>{if(!e||e.length!==1)throw new Error("Softmax op requires 1 input.")},Dg=(e,r)=>{let t=e.inputs[0],s=t.dims,o=Pe.size(s),n=s.length,a=Pe.normalizeAxis(r.axis,n),i=a<s.length-1,l,c=[];i?(c=Array.from({length:n},(v,P)=>P),c[a]=n-1,c[n-1]=a,l=e.compute(rs(t,c),{inputs:[t],outputs:[-1]})[0]):l=t;let p=l.dims,d=p[n-1],u=o/d,_=or(d),f=d/_,b=64;u===1&&(b=256);let A=(v,P)=>P===4?`max(max(${v}.x, ${v}.y), max(${v}.z, ${v}.w))`:P===2?`max(${v}.x, ${v}.y)`:P===3?`max(max(${v}.x, ${v}.y), ${v}.z)`:v,g=Oe("x",l.dataType,l.dims,_),y=at("result",l.dataType,l.dims,_),C=g.type.value,x=Sr(l.dataType)==="f32"?`var threadMax = ${C}(-3.402823e+38f);`:`var threadMax = ${C}(-65504.0h);`,M=v=>`
      var<workgroup> rowMaxShared : ${C};
      var<workgroup> rowSumShared : ${C};
      var<workgroup> threadShared : array<${C}, ${b}>;

      fn getValue(row: i32, col: i32, row_stride: i32) -> ${C} {
        let index = row * row_stride + col;
        return x[index];
      }

      fn setValue(row: i32, col: i32, row_stride: i32, value: ${C}) {
        let index = row * row_stride + col;
        result[index] = value;
      }
      ${v.registerUniform("packedCols","i32").declareVariables(g,y)}
      ${v.mainStart(b)}
        let gindex = i32(global_idx);
        let lindex = i32(local_idx);
        const wg = ${b};
        let row = gindex / wg;
        let cols = uniforms.packedCols;
        let row_stride : i32 = uniforms.packedCols;

        // find the rows max
        ${x}
        for (var col = lindex; col < cols; col += wg) {
          let value = getValue(row, col, row_stride);
          threadMax = max(threadMax, value);
        }
        if (lindex < cols) {
          threadShared[lindex] = threadMax;
        }
        workgroupBarrier();

        var reduceSize = min(cols, wg);
        for (var currSize = reduceSize >> 1;  currSize > 0; currSize = reduceSize >> 1) {
          reduceSize = currSize + (reduceSize & 1);
          if (lindex < currSize) {
            threadShared[lindex] = max(threadShared[lindex], threadShared[lindex + reduceSize]);
          }
          workgroupBarrier();
        }
        if (lindex == 0) {
          rowMaxShared = ${C}(${A("threadShared[0]",_)});
        }
        workgroupBarrier();

        // find the rows sum
        var threadSum = ${C}(0.0);
        for (var col = lindex; col < cols; col += wg) {
          let subExp = exp(getValue(row, col, row_stride) - rowMaxShared);
          threadSum += subExp;
        }
        threadShared[lindex] = threadSum;
        workgroupBarrier();

        for (var currSize = wg >> 1;  currSize > 0; currSize = currSize >> 1) {
          if (lindex < currSize) {
            threadShared[lindex] = threadShared[lindex] + threadShared[lindex + currSize];
          }
          workgroupBarrier();
        }
        if (lindex == 0) {
          rowSumShared = ${C}(${Js("threadShared[0]",_)});
        }
        workgroupBarrier();

        // calculate final value for each element in the row
        for (var col = lindex; col < cols; col += wg) {
          let value = exp(getValue(row, col, row_stride) - rowMaxShared) / rowSumShared;
          setValue(row, col, row_stride, value);
        }
      }`,T=e.compute({name:"Softmax",shaderCache:{hint:`${_};${b}`,inputDependencies:["type"]},getRunData:()=>({outputs:[{dims:p,dataType:l.dataType}],dispatchGroup:{x:u},programUniforms:[{type:6,data:f}]}),getShaderSource:M},{inputs:[l],outputs:[i?-1:0]})[0];i&&e.compute(rs(T,c),{inputs:[T]})},Lg=(e,r)=>{Og(e.inputs),Dg(e,r)},zg=e=>jt({axis:e.axis})}),au,Rg,Bg,Ng,jg,FT=Ue(()=>{Mt(),Pt(),Ct(),au=e=>Array.from(e.getBigInt64Array(),Number),Rg=e=>{if(!e||e.length!==2)throw new Error("Tile requires 2 inputs.");if(e[0].dataType!==1&&e[0].dataType!==10&&e[0].dataType!==6&&e[0].dataType!==12)throw new Error("Tile only support float, float16, int32, and uint32 data types");if(e[1].dataType!==7)throw new Error("Tile `repeats` input should be of int64 data type");if(e[1].dims.length!==1)throw new Error("Tile `repeats` input should be 1-D");if(au(e[1]).length!==e[0].dims.length)throw new Error("Tile `repeats` input should have same number of elements as rank of input data tensor")},Bg=(e,r)=>{let t=[];for(let s=0;s<e.length;++s)t.push(e[s]*r[s]);return t},Ng=(e,r)=>{let t=e[0].dims,s=r??au(e[1]),o=Bg(t,s),n=Pe.size(o),a=e[0].dataType,i=Oe("input",a,t.length),l=at("output",a,o.length),c=p=>`
      const inputShape = ${i.indices(...t)};
      ${p.registerUniform("output_size","u32").declareVariables(i,l)}
      ${p.mainStart()}
      ${p.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.output_size")}
      let output_indices = ${l.offsetToIndices("global_idx")};
      var input_indices: ${i.type.indices};
      for (var i = 0; i < ${t.length}; i++) {
        let input_dim_i = ${i.indicesGet("uniforms.input_shape","i")};
        let input_dim_value = ${l.indicesGet("output_indices","i")}  % input_dim_i;

        ${i.indicesSet("input_indices","i","input_dim_value")}
      }
      ${l.setByOffset("global_idx",i.getByIndices("input_indices"))}
    }`;return{name:"Tile",shaderCache:{hint:`${s}`,inputDependencies:["rank"]},getRunData:()=>({outputs:[{dims:o,dataType:e[0].dataType}],dispatchGroup:{x:Math.ceil(n/64)},programUniforms:[{type:12,data:n},...lt(e[0].dims,o)]}),getShaderSource:c}},jg=e=>{Rg(e.inputs),e.compute(Ng(e.inputs),{inputs:[0]})}}),Vg,Ug,Wg,OT=Ue(()=>{Mt(),Pt(),Ct(),Vg=(e,r,t,s,o)=>{let n=at("output_data",o,t.length,4),a=Oe("a_data",r[1].dataType,r[1].dims.length,4),i=Oe("b_data",r[2].dataType,r[2].dims.length,4),l=Oe("c_data",r[0].dataType,r[0].dims.length,4),c,p=(d,u,_)=>`select(${u}, ${d}, ${_})`;if(!s)c=n.setByOffset("global_idx",p(a.getByOffset("global_idx"),i.getByOffset("global_idx"),l.getByOffset("global_idx")));else{let d=(u,_,f="")=>{let b=`a_data[index_a${_}][component_a${_}]`,A=`b_data[index_b${_}][component_b${_}]`,g=`bool(c_data[index_c${_}] & (0xffu << (component_c${_} * 8)))`;return`
            let output_indices${_} = ${n.offsetToIndices(`global_idx * 4u + ${_}u`)};
            let offset_a${_} = ${a.broadcastedIndicesToOffset(`output_indices${_}`,n)};
            let offset_b${_} = ${i.broadcastedIndicesToOffset(`output_indices${_}`,n)};
            let offset_c${_} = ${l.broadcastedIndicesToOffset(`output_indices${_}`,n)};
            let index_a${_} = offset_a${_} / 4u;
            let index_b${_} = offset_b${_} / 4u;
            let index_c${_} = offset_c${_} / 4u;
            let component_a${_} = offset_a${_} % 4u;
            let component_b${_} = offset_b${_} % 4u;
            let component_c${_} = offset_c${_} % 4u;
            ${u}[${_}] = ${f}(${p(b,A,g)});
          `};o===9?c=`
            var data = vec4<u32>(0);
            ${d("data",0,"u32")}
            ${d("data",1,"u32")}
            ${d("data",2,"u32")}
            ${d("data",3,"u32")}
            output_data[global_idx] = dot(vec4<u32>(0x1, 0x100, 0x10000, 0x1000000), vec4<u32>(data));`:c=`
            ${d("output_data[global_idx]",0)}
            ${d("output_data[global_idx]",1)}
            ${d("output_data[global_idx]",2)}
            ${d("output_data[global_idx]",3)}
          `}return`
        ${e.registerUniform("vec_size","u32").declareVariables(l,a,i,n)}
        ${e.mainStart()}
        ${e.guardAgainstOutOfBoundsWorkgroupSizes("uniforms.vec_size")}
        ${c}
      }`},Ug=e=>{let r=e[1].dims,t=e[2].dims,s=e[0].dims,o=e[1].dataType,n=!(Pe.areEqual(r,t)&&Pe.areEqual(t,s)),a=r,i=Pe.size(r);if(n){let c=Gn.calcShape(Gn.calcShape(r,t,!1),s,!1);if(!c)throw new Error("Can't perform where op on the given tensors");a=c,i=Pe.size(a)}let l=Math.ceil(i/4);return{name:"Where",shaderCache:{inputDependencies:["rank","rank","rank"]},getShaderSource:c=>Vg(c,e,a,n,o),getRunData:()=>({outputs:[{dims:a,dataType:o}],dispatchGroup:{x:Math.ceil(i/64/4)},programUniforms:[{type:12,data:l},...lt(s,r,t,a)]})}},Wg=e=>{e.compute(Ug(e.inputs))}}),Gg,DT=Ue(()=>{Qx(),bl(),Xx(),Jx(),Yx(),Zx(),eT(),oT(),iT(),lT(),uT(),cT(),dT(),pT(),hT(),mT(),fT(),_T(),gT(),wT(),MT(),bT(),yT(),vT(),xT(),s_(),TT(),ET(),PT(),CT(),ST(),gl(),$T(),h_(),AT(),kT(),IT(),c_(),FT(),Ys(),Tl(),OT(),Gg=new Map([["Abs",[Th]],["Acos",[Eh]],["Acosh",[Ph]],["Add",[hm]],["ArgMax",[uh,Ml]],["ArgMin",[lh,Ml]],["Asin",[Ch]],["Asinh",[Sh]],["Atan",[$h]],["Atanh",[Ah]],["Attention",[fh]],["AveragePool",[U_,V_]],["BatchNormalization",[Mh]],["BiasAdd",[vh]],["BiasSplitGelu",[cm]],["Cast",[Ih,kh]],["Ceil",[Dh]],["Clip",[Oh]],["Concat",[Cm,Sm]],["Conv",[zl,Dl]],["ConvTranspose",[tf,Ym]],["Cos",[Lh]],["Cosh",[zh]],["CumSum",[sf,nf]],["DepthToSpace",[uf,cf]],["DequantizeLinear",[Y_,Z_]],["Div",[mm]],["Einsum",[_f,gf]],["Elu",[Rh,Mo]],["Equal",[fm]],["Erf",[Bh]],["Exp",[Nh]],["Expand",[yf]],["FastGelu",[xf]],["Floor",[jh]],["FusedConv",[zl,Dl]],["Gather",[Cf,Pf]],["GatherElements",[Rf,zf]],["GatherBlockQuantized",[Ff,Of]],["GatherND",[$f,Af]],["Gelu",[Vh]],["Gemm",[Vf,jf]],["GlobalAveragePool",[G_,W_]],["GlobalMaxPool",[Q_,q_]],["Greater",[Mm]],["GreaterOrEqual",[ym]],["GridSample",[Jf,Yf]],["GroupQueryAttention",[g_]],["HardSigmoid",[Xh,Qh]],["InstanceNormalization",[b_]],["LayerNormalization",[x_]],["LeakyRelu",[Uh,Mo]],["Less",[bm]],["LessOrEqual",[vm]],["Log",[nm]],["MatMul",[E_]],["MatMulNBits",[$_,A_]],["MaxPool",[H_,K_]],["Mul",[_m]],["MultiHeadAttention",[r_,e_]],["Neg",[Gh]],["Not",[Wh]],["Pad",[B_]],["Pow",[gm]],["QuickGelu",[im,Mo]],["Range",[rg]],["Reciprocal",[Hh]],["ReduceMin",[sh]],["ReduceMean",[Yp]],["ReduceMax",[rh]],["ReduceSum",[oh]],["ReduceProd",[nh]],["ReduceL1",[Zp]],["ReduceL2",[eh]],["ReduceLogSum",[ih]],["ReduceLogSumExp",[th]],["ReduceSumSquare",[ah]],["Relu",[Kh]],["Resize",[xg,Tg]],["RotaryEmbedding",[p_]],["ScatterND",[ag,og]],["Sigmoid",[qh]],["Sin",[Jh]],["Sinh",[Yh]],["Slice",[Ig,Fg]],["SkipLayerNormalization",[Cg]],["Split",[l_,u_]],["Sqrt",[Zh]],["Softmax",[Lg,zg]],["Sub",[wm]],["Tan",[em]],["Tanh",[tm]],["ThresholdedRelu",[sm,Mo]],["Tile",[jg]],["Transpose",[Mp,bp]],["Where",[Wg]]])}),Hg,LT=Ue(()=>{bs(),js(),Ct(),Hg=class{constructor(e){this.backend=e,this.repo=new Map,this.attributesBound=!1}getArtifact(e){return this.repo.get(e)}setArtifact(e,r){this.repo.set(e,r)}run(e,r,t,s,o){Ms(e.programInfo.name);let n=this.backend.device,a=this.backend.getComputePassEncoder();this.backend.writeTimestamp(this.backend.pendingDispatchNumber*2);let i=[];for(let c of r)i.push({binding:i.length,resource:{buffer:c.buffer}});for(let c of t)i.push({binding:i.length,resource:{buffer:c.buffer}});o&&i.push({binding:i.length,resource:o});let l=n.createBindGroup({layout:e.computePipeline.getBindGroupLayout(0),entries:i,label:e.programInfo.name});if(this.backend.sessionStatus==="capturing"){let c={kernelId:this.backend.currentKernelId,computePipeline:e.computePipeline,bindGroup:l,dispatchGroup:s};this.backend.capturedCommandList.get(this.backend.currentSessionId).push(c)}a.setPipeline(e.computePipeline),a.setBindGroup(0,l),a.dispatchWorkgroups(...s),this.backend.writeTimestamp(this.backend.pendingDispatchNumber*2+1),this.backend.pendingDispatchNumber++,(this.backend.pendingDispatchNumber>=this.backend.maxDispatchNumber||this.backend.queryType==="at-passes")&&this.backend.endComputePass(),this.backend.pendingDispatchNumber>=this.backend.maxDispatchNumber&&this.backend.flush(),us(e.programInfo.name)}dispose(){}build(e,r){Ms(e.name);let t=this.backend.device,s=[];[{feature:"shader-f16",extension:"f16"},{feature:"subgroups",extension:"subgroups"}].forEach(c=>{t.features.has(c.feature)&&s.push(`enable ${c.extension};`)});let o=hp(r,this.backend.device.limits),n=e.getShaderSource(o),a=`${s.join(`
`)}
${o.additionalImplementations}
${n}`,i=t.createShaderModule({code:a,label:e.name});Dt("verbose",()=>`[WebGPU] ${e.name} shader code: ${a}`);let l=t.createComputePipeline({compute:{module:i,entryPoint:"main"},layout:"auto",label:e.name});return us(e.name),{programInfo:e,computePipeline:l,uniformVariablesInfo:o.variablesInfo}}normalizeDispatchGroupSize(e){let r=typeof e=="number"?e:e.x,t=typeof e=="number"?1:e.y||1,s=typeof e=="number"?1:e.z||1,o=this.backend.device.limits.maxComputeWorkgroupsPerDimension;if(r<=o&&t<=o&&s<=o)return[r,t,s];let n=r*t*s,a=Math.ceil(Math.sqrt(n));if(a>o){if(a=Math.ceil(Math.cbrt(n)),a>o)throw new Error("Total dispatch size exceeds WebGPU maximum.");return[a,a,a]}else return[a,a,1]}}}),Kg={};Un(Kg,{WebGpuBackend:()=>Jg});var qg,Qg,Xg,Jg,zT=Ue(()=>{bs(),Mt(),js(),Zd(),Kx(),DT(),LT(),qg=(e,r)=>{if(r.length!==e.length)throw new Error(`inputDependencies length ${r.length} is not equal to inputTensors length ${e.length}.`);let t=[];for(let s=0;s<e.length;++s){let o=e[s].dataType;switch(r[s]){case"none":{t.push("");break}case"type":{t.push(`${o}`);break}case"rank":{let n=e[s].dims.length;t.push(`${o};${n}`);break}case"dims":{let n=e[s].dims.join(",");t.push(`${o};${n}`);break}default:throw new Error(`unsupported input dependency: ${r[s]}`)}}return t.join("|")},Qg=(e,r,t)=>{var o,n;let s=e.name;return(o=e.shaderCache)!=null&&o.hint&&(s+="["+e.shaderCache.hint+"]"),s+=":"+t+`:${qg(r,((n=e.shaderCache)==null?void 0:n.inputDependencies)??new Array(r.length).fill("dims"))}`,s},Xg=class{constructor(e){e&&(this.architecture=e.architecture,this.vendor=e.vendor)}isArchitecture(e){return this.architecture===e}isVendor(e){return this.vendor===e}},Jg=class{constructor(){this.currentSessionId=null,this.currentKernelId=null,this.commandEncoder=null,this.computePassEncoder=null,this.maxDispatchNumber=16,this.pendingDispatchNumber=0,this.pendingKernels=[],this.pendingQueries=new Map,this.sessionStatus="default",this.capturedCommandList=new Map,this.capturedPendingKernels=new Map,this.sessionExternalDataMapping=new Map}get currentKernelCustomData(){if(this.currentKernelId===null)throw new Error("currentKernelCustomData(): currentKernelId is null. (should not happen)");let e=this.kernelCustomData.get(this.currentKernelId);return e||(e={},this.kernelCustomData.set(this.currentKernelId,e)),e}async initialize(e,r){this.env=e;let t=[],s={requiredLimits:{maxComputeWorkgroupStorageSize:r.limits.maxComputeWorkgroupStorageSize,maxComputeWorkgroupsPerDimension:r.limits.maxComputeWorkgroupsPerDimension,maxStorageBufferBindingSize:r.limits.maxStorageBufferBindingSize,maxBufferSize:r.limits.maxBufferSize,maxComputeInvocationsPerWorkgroup:r.limits.maxComputeInvocationsPerWorkgroup,maxComputeWorkgroupSizeX:r.limits.maxComputeWorkgroupSizeX,maxComputeWorkgroupSizeY:r.limits.maxComputeWorkgroupSizeY,maxComputeWorkgroupSizeZ:r.limits.maxComputeWorkgroupSizeZ},requiredFeatures:t},o=n=>r.features.has(n)&&t.push(n)&&!0;o("chromium-experimental-timestamp-query-inside-passes")||o("timestamp-query"),o("shader-f16"),o("subgroups"),this.device=await r.requestDevice(s),this.adapterInfo=new Xg(r.info||await r.requestAdapterInfo()),this.gpuDataManager=up(this),this.programManager=new Hg(this),this.kernels=new Map,this.kernelPersistentData=new Map,this.kernelCustomData=new Map,tl(e.logLevel,!!e.debug),this.device.onuncapturederror=n=>{n.error instanceof GPUValidationError&&console.error(`An uncaught WebGPU validation error was raised: ${n.error.message}`)},Object.defineProperty(this.env.webgpu,"device",{value:this.device,writable:!1,enumerable:!0,configurable:!1}),Object.defineProperty(this.env.webgpu,"adapter",{value:r,writable:!1,enumerable:!0,configurable:!1}),this.setQueryType()}dispose(){typeof this.querySet<"u"&&this.querySet.destroy(),this.gpuDataManager.dispose()}getCommandEncoder(){return this.commandEncoder||(this.commandEncoder=this.device.createCommandEncoder()),this.commandEncoder}getComputePassEncoder(){if(!this.computePassEncoder){let e=this.getCommandEncoder(),r={};this.queryType==="at-passes"&&(r.timestampWrites={querySet:this.querySet,beginningOfPassWriteIndex:this.pendingDispatchNumber*2,endOfPassWriteIndex:this.pendingDispatchNumber*2+1}),this.computePassEncoder=e.beginComputePass(r)}return this.computePassEncoder}endComputePass(){this.computePassEncoder&&(this.computePassEncoder.end(),this.computePassEncoder=null)}flush(){if(!this.commandEncoder)return;Ms(),this.endComputePass();let e;this.queryType!=="none"&&(this.commandEncoder.resolveQuerySet(this.querySet,0,this.pendingDispatchNumber*2,this.queryResolveBuffer,0),e=this.device.createBuffer({size:this.pendingDispatchNumber*2*8,usage:GPUBufferUsage.MAP_READ|GPUBufferUsage.COPY_DST}),this.pendingQueries.set(e,this.pendingKernels),this.pendingKernels=[],this.commandEncoder.copyBufferToBuffer(this.queryResolveBuffer,0,e,0,this.pendingDispatchNumber*2*8)),this.device.queue.submit([this.commandEncoder.finish()]),this.gpuDataManager.refreshPendingBuffers(),this.commandEncoder=null,this.pendingDispatchNumber=0,this.queryType!=="none"&&e.mapAsync(GPUMapMode.READ).then(()=>{var s;let r=new BigUint64Array(e.getMappedRange()),t=this.pendingQueries.get(e);for(let o=0;o<r.length/2;o++){let n=t[o],a=n.kernelId,i=this.kernels.get(a),l=i.kernelType,c=i.kernelName,p=n.programName,d=n.inputTensorViews,u=n.outputTensorViews,_=r[o*2],f=r[o*2+1];typeof this.queryTimeBase>"u"&&(this.queryTimeBase=_);let b=Number(_-this.queryTimeBase),A=Number(f-this.queryTimeBase);if(!Number.isSafeInteger(b)||!Number.isSafeInteger(A))throw new RangeError("incorrect timestamp range");if((s=this.env.webgpu.profiling)!=null&&s.ondata)this.env.webgpu.profiling.ondata({version:1,inputsMetadata:d.map(g=>({dims:g.dims,dataType:Ns(g.dataType)})),outputsMetadata:u.map(g=>({dims:g.dims,dataType:Ns(g.dataType)})),kernelId:a,kernelType:l,kernelName:c,programName:p,startTime:b,endTime:A});else{let g="";d.forEach((C,x)=>{g+=`input[${x}]: [${C.dims}] | ${Ns(C.dataType)}, `});let y="";u.forEach((C,x)=>{y+=`output[${x}]: [${C.dims}] | ${Ns(C.dataType)}, `}),console.log(`[profiling] kernel "${a}|${l}|${c}|${p}" ${g}${y}execution time: ${A-b} ns`)}mo("GPU",`${p}::${_}::${f}`)}e.unmap(),this.pendingQueries.delete(e)}),us()}run(e,r,t,s,o,n){Ms(e.name);let a=[];for(let y=0;y<r.length;++y){let C=r[y].data;if(C===0)continue;let x=this.gpuDataManager.get(C);if(!x)throw new Error(`no GPU data for input: ${C}`);a.push(x)}let{outputs:i,dispatchGroup:l,programUniforms:c}=e.getRunData(r),p=t.length===0?i.map((y,C)=>C):t;if(p.length!==i.length)throw new Error(`Output size ${p.length} must be equal to ${i.length}.`);let d=[],u=[];for(let y=0;y<i.length;++y){if(!Number.isInteger(p[y])||p[y]<-3||p[y]>=n)throw new Error(`Invalid output index: ${p[y]}`);if(p[y]===-3)continue;let C=p[y]===-1,x=p[y]===-2,M=C||x?o(i[y].dataType,i[y].dims):s(p[y],i[y].dataType,i[y].dims);if(d.push(M),M.data===0)continue;let T=this.gpuDataManager.get(M.data);if(!T)throw new Error(`no GPU data for output: ${M.data}`);if(C&&this.temporaryData.push(T),x){let v=this.kernelPersistentData.get(this.currentKernelId);v||(v=[],this.kernelPersistentData.set(this.currentKernelId,v)),v.push(T)}u.push(T)}if(a.length!==r.length||u.length!==d.length){if(u.length===0)return us(e.name),d;throw new Error(`Program ${e.name} has zero-sized tensor(s) in inputs or outputs. This is not supported now.`)}let _;if(c){let y=0,C=[];c.forEach(v=>{let P=typeof v.data=="number"?[v.data]:v.data;if(P.length===0)return;let F=v.type===10?2:4,D,K;v.type===10?(K=P.length>4?16:P.length>2?8:P.length*F,D=P.length>4?16:F*P.length):(K=P.length<=2?P.length*F:16,D=16),y=Math.ceil(y/K)*K,C.push(y);let U=v.type===10?8:4;y+=P.length>4?Math.ceil(P.length/U)*D:P.length*F});let x=16;y=Math.ceil(y/x)*x;let M=new ArrayBuffer(y);c.forEach((v,P)=>{let F=C[P],D=typeof v.data=="number"?[v.data]:v.data;if(v.type===6)new Int32Array(M,F,D.length).set(D);else if(v.type===12)new Uint32Array(M,F,D.length).set(D);else if(v.type===10)new Uint16Array(M,F,D.length).set(D);else if(v.type===1)new Float32Array(M,F,D.length).set(D);else throw new Error(`Unsupported uniform type: ${Ns(v.type)}`)});let T=this.gpuDataManager.create(y,GPUBufferUsage.COPY_DST|GPUBufferUsage.UNIFORM);this.device.queue.writeBuffer(T.buffer,0,M,0,y),this.gpuDataManager.release(T.id),_={offset:0,size:y,buffer:T.buffer}}let f=this.programManager.normalizeDispatchGroupSize(l),b=f[1]===1&&f[2]===1,A=Qg(e,r,b),g=this.programManager.getArtifact(A);if(g||(g=this.programManager.build(e,f),this.programManager.setArtifact(A,g),Dt("info",()=>`[artifact] key: ${A}, programName: ${e.name}`)),c&&g.uniformVariablesInfo){if(c.length!==g.uniformVariablesInfo.length)throw new Error(`Uniform variables count mismatch: expect ${g.uniformVariablesInfo.length}, got ${c.length} in program "${g.programInfo.name}".`);for(let y=0;y<c.length;y++){let C=c[y],x=C.type,M=typeof C.data=="number"?1:C.data.length,[T,v]=g.uniformVariablesInfo[y];if(x!==T||M!==v)throw new Error(`Uniform variable ${y} mismatch: expect type ${T} with size ${v}, got type ${x} with size ${M} in program "${g.programInfo.name}".`)}}if(Dt("info",()=>`[ProgramManager] run "${e.name}" (key=${A}) with ${f[0]}x${f[1]}x${f[2]}`),this.queryType!=="none"||this.sessionStatus==="capturing"){let y={kernelId:this.currentKernelId,programName:g.programInfo.name,inputTensorViews:r,outputTensorViews:d};this.pendingKernels.push(y),this.sessionStatus==="capturing"&&this.capturedPendingKernels.get(this.currentSessionId).push(y)}return this.programManager.run(g,a,u,f,_),us(e.name),d}upload(e,r){this.gpuDataManager.upload(e,r)}memcpy(e,r){this.gpuDataManager.memcpy(e,r)}async download(e,r){await this.gpuDataManager.download(e,r)}alloc(e){return this.gpuDataManager.create(e).id}free(e){return this.gpuDataManager.release(e)}createKernel(e,r,t,s){let o=Gg.get(e);if(!o)throw new Error(`kernel not implemented: ${e}`);let n={kernelType:e,kernelName:s,kernelEntry:o[0],attributes:[o[1],t]};this.kernels.set(r,n)}releaseKernel(e){let r=this.kernelPersistentData.get(e);if(r){for(let t of r)this.gpuDataManager.release(t.id);this.kernelPersistentData.delete(e)}this.kernelCustomData.delete(e),this.kernels.delete(e)}computeKernel(e,r,t){let s=this.kernels.get(e);if(!s)throw new Error(`kernel not created: ${e}`);let o=s.kernelType,n=s.kernelName,a=s.kernelEntry,i=s.attributes;if(this.currentKernelId!==null)throw new Error(`kernel "[${o}] ${n}" is not allowed to be called recursively`);this.currentKernelId=e,i[0]&&(i[1]=i[0](i[1]),i[0]=void 0),Dt("info",()=>`[WebGPU] Start to run kernel "[${o}] ${n}"...`);let l=this.env.debug;this.temporaryData=[];try{return l&&this.device.pushErrorScope("validation"),a(r,i[1]),0}catch(c){return t.push(Promise.resolve(`[WebGPU] Kernel "[${o}] ${n}" failed. ${c}`)),1}finally{l&&t.push(this.device.popErrorScope().then(c=>c?`GPU validation error for kernel "[${o}] ${n}": ${c.message}`:null));for(let c of this.temporaryData)this.gpuDataManager.release(c.id);this.temporaryData=[],this.currentKernelId=null}}registerBuffer(e,r,t,s){let o=this.sessionExternalDataMapping.get(e);o||(o=new Map,this.sessionExternalDataMapping.set(e,o));let n=o.get(r),a=this.gpuDataManager.registerExternalBuffer(t,s,n);return o.set(r,[a,t]),a}unregisterBuffers(e){let r=this.sessionExternalDataMapping.get(e);r&&(r.forEach(t=>this.gpuDataManager.unregisterExternalBuffer(t[0])),this.sessionExternalDataMapping.delete(e))}getBuffer(e){let r=this.gpuDataManager.get(e);if(!r)throw new Error(`no GPU data for buffer: ${e}`);return r.buffer}createDownloader(e,r,t){return async()=>{let s=await pl(this,e,r);return rl(s.buffer,t)}}writeTimestamp(e){this.queryType==="inside-passes"&&this.computePassEncoder.writeTimestamp(this.querySet,e)}setQueryType(){var e;this.queryType="none",(((e=this.env.webgpu.profiling)==null?void 0:e.mode)==="default"||(typeof this.env.trace>"u"?this.env.wasm.trace:this.env.trace))&&(this.device.features.has("chromium-experimental-timestamp-query-inside-passes")?this.queryType="inside-passes":this.device.features.has("timestamp-query")&&(this.queryType="at-passes"),this.queryType!=="none"&&typeof this.querySet>"u"&&(this.querySet=this.device.createQuerySet({type:"timestamp",count:this.maxDispatchNumber*2}),this.queryResolveBuffer=this.device.createBuffer({size:this.maxDispatchNumber*2*8,usage:GPUBufferUsage.COPY_SRC|GPUBufferUsage.QUERY_RESOLVE})))}captureBegin(){Dt("info","captureBegin"),this.capturedCommandList.get(this.currentSessionId)||this.capturedCommandList.set(this.currentSessionId,[]),this.capturedPendingKernels.get(this.currentSessionId)||this.capturedPendingKernels.set(this.currentSessionId,[]),this.flush(),this.sessionStatus="capturing"}captureEnd(){Dt("info","captureEnd"),this.flush(),this.sessionStatus="default"}replay(){Dt("info","replay"),this.sessionStatus="replaying";let e=this.capturedCommandList.get(this.currentSessionId),r=this.capturedPendingKernels.get(this.currentSessionId),t=e.length;this.pendingKernels=[];for(let s=0;s<t;s++){let o=this.getComputePassEncoder(),n=e[s];this.writeTimestamp(this.pendingDispatchNumber*2),o.setPipeline(n.computePipeline),o.setBindGroup(0,n.bindGroup),o.dispatchWorkgroups(...n.dispatchGroup),this.writeTimestamp(this.pendingDispatchNumber*2+1),this.pendingDispatchNumber++,this.queryType!=="none"&&this.pendingKernels.push(r[s]),(this.pendingDispatchNumber>=this.maxDispatchNumber||this.queryType==="at-passes")&&this.endComputePass(),this.pendingDispatchNumber>=this.maxDispatchNumber&&this.flush()}this.flush(),this.sessionStatus="default"}onCreateSession(){this.gpuDataManager.onCreateSession()}onReleaseSession(e){this.unregisterBuffers(e),this.capturedCommandList.has(e)&&this.capturedCommandList.delete(e),this.capturedPendingKernels.has(e)&&this.capturedPendingKernels.delete(e),this.gpuDataManager.onReleaseSession(e)}onRunStart(e){this.currentSessionId=e,this.setQueryType()}}}),Yg={};Un(Yg,{init:()=>ew});var ya,Zg,ew,RT=Ue(()=>{Mt(),js(),Pt(),Hx(),ya=class ix{constructor(r,t,s,o){this.module=r,this.dataType=t,this.data=s,this.dims=o}getFloat32Array(){if(this.dataType!==1)throw new Error("Invalid data type");let r=Pe.size(this.dims);return r===0?new Float32Array:new Float32Array(this.module.HEAP8.buffer,this.data,r)}getBigInt64Array(){if(this.dataType!==7)throw new Error("Invalid data type");let r=Pe.size(this.dims);return r===0?new BigInt64Array:new BigInt64Array(this.module.HEAP8.buffer,this.data,r)}getInt32Array(){if(this.dataType!==6)throw new Error("Invalid data type");let r=Pe.size(this.dims);return r===0?new Int32Array:new Int32Array(this.module.HEAP8.buffer,this.data,r)}getUint16Array(){if(this.dataType!==10&&this.dataType!==4)throw new Error("Invalid data type");let r=Pe.size(this.dims);return r===0?new Uint16Array:new Uint16Array(this.module.HEAP8.buffer,this.data,r)}reshape(r){if(Pe.size(r)!==Pe.size(this.dims))throw new Error("Invalid new shape");return new ix(this.module,this.dataType,this.data,r)}},Zg=class{constructor(e,r,t){this.module=e,this.backend=r,this.customDataOffset=0,this.customDataSize=0,this.adapterInfo=r.adapterInfo;let s=e.PTR_SIZE,o=t/e.PTR_SIZE,n=s===4?"i32":"i64";this.opKernelContext=Number(e.getValue(s*o++,n));let a=Number(e.getValue(s*o++,n));this.outputCount=Number(e.getValue(s*o++,n)),this.customDataOffset=Number(e.getValue(s*o++,"*")),this.customDataSize=Number(e.getValue(s*o++,n));let i=[];for(let l=0;l<a;l++){let c=Number(e.getValue(s*o++,n)),p=Number(e.getValue(s*o++,"*")),d=Number(e.getValue(s*o++,n)),u=[];for(let _=0;_<d;_++)u.push(Number(e.getValue(s*o++,n)));i.push(new ya(e,c,p,u))}this.inputs=i}get kernelCustomData(){return this.backend.currentKernelCustomData}get customDataBuffer(){return this.module.HEAPU8.subarray(this.customDataOffset,this.customDataOffset+this.customDataSize)}compute(e,r){var a;let t=((a=r==null?void 0:r.inputs)==null?void 0:a.map(i=>typeof i=="number"?this.inputs[i]:i))??this.inputs,s=(r==null?void 0:r.outputs)??[],o=(i,l,c)=>new ya(this.module,l,this.output(i,c),c),n=(i,l)=>{let c=fn(i,l);if(!c)throw new Error(`Unsupported data type: ${i}`);let p=c>0?this.backend.gpuDataManager.create(c).id:0;return new ya(this.module,i,p,l)};return this.backend.run(e,t,s,o,n,this.outputCount)}output(e,r){let t=this.module.stackSave();try{let s=this.module.PTR_SIZE,o=s===4?"i32":"i64",n=this.module.stackAlloc((1+r.length)*s);this.module.setValue(n,r.length,o);for(let a=0;a<r.length;a++)this.module.setValue(n+s*(a+1),r[a],o);return this.module._JsepOutput(this.opKernelContext,e,n)}catch(s){throw new Error(`Failed to generate kernel's output[${e}] with dims [${r}]. If you are running with pre-allocated output, please make sure the output type/dims are correct. Error: ${s}`)}finally{this.module.stackRestore(t)}}},ew=async(e,r,t,s)=>{let o=r.jsepInit;if(!o)throw new Error("Failed to initialize JSEP. The WebAssembly module is not built with JSEP support.");if(e==="webgpu"){let n=(zT(),co(Kg)).WebGpuBackend,a=new n;await a.initialize(t,s),o("webgpu",[a,i=>a.alloc(Number(i)),i=>a.free(i),(i,l,c,p=!1)=>{if(p)Dt("verbose",()=>`[WebGPU] jsepCopyGpuToGpu: src=${Number(i)}, dst=${Number(l)}, size=${Number(c)}`),a.memcpy(Number(i),Number(l));else{Dt("verbose",()=>`[WebGPU] jsepCopyCpuToGpu: dataOffset=${Number(i)}, gpuDataId=${Number(l)}, size=${Number(c)}`);let d=r.HEAPU8.subarray(Number(i>>>0),Number(i>>>0)+Number(c));a.upload(Number(l),d)}},async(i,l,c)=>{Dt("verbose",()=>`[WebGPU] jsepCopyGpuToCpu: gpuDataId=${i}, dataOffset=${l}, size=${c}`),await a.download(Number(i),()=>r.HEAPU8.subarray(Number(l)>>>0,Number(l+c)>>>0))},(i,l,c)=>a.createKernel(i,Number(l),c,r.UTF8ToString(r._JsepGetNodeName(Number(l)))),i=>a.releaseKernel(i),(i,l,c,p)=>{Dt("verbose",()=>`[WebGPU] jsepRun: sessionHandle=${c}, kernel=${i}, contextDataOffset=${l}`);let d=new Zg(r,a,Number(l));return a.computeKernel(Number(i),d,p)},()=>a.captureBegin(),()=>a.captureEnd(),()=>a.replay()])}else{let n=new op(t);o("webnn",[n,()=>n.reserveTensorId(),a=>n.releaseTensorId(a),async(a,i,l,c,p)=>n.ensureTensor(a,i,l,c,p),(a,i)=>{n.uploadTensor(a,i)},async(a,i)=>n.downloadTensor(a,i)])}}}),tw,iu,lu,Zs,rw,uu,va,cu,du,pu,hu,mu,fu,sw=Ue(()=>{Ux(),Wx(),Mt(),mn(),Qi(),Ud(),tw=(e,r)=>{Xt()._OrtInit(e,r)!==0&&Wt("Can't initialize onnxruntime.")},iu=async e=>{tw(e.wasm.numThreads,ia(e.logLevel))},lu=async(e,r)=>{var t,s;(s=(t=Xt()).asyncInit)==null||s.call(t);{let o=(RT(),co(Yg)).init;if(r==="webgpu"){if(typeof navigator>"u"||!navigator.gpu)throw new Error("WebGPU is not supported in current environment");let n=e.webgpu.adapter;if(n){if(typeof n.limits!="object"||typeof n.features!="object"||typeof n.requestDevice!="function")throw new Error("Invalid GPU adapter set in `env.webgpu.adapter`. It must be a GPUAdapter object.")}else{let a=e.webgpu.powerPreference;if(a!==void 0&&a!=="low-power"&&a!=="high-performance")throw new Error(`Invalid powerPreference setting: "${a}"`);let i=e.webgpu.forceFallbackAdapter;if(i!==void 0&&typeof i!="boolean")throw new Error(`Invalid forceFallbackAdapter setting: "${i}"`);if(n=await navigator.gpu.requestAdapter({powerPreference:a,forceFallbackAdapter:i}),!n)throw new Error('Failed to get GPU adapter. You may need to enable flag "--enable-unsafe-webgpu" if you are using Chrome.')}await o("webgpu",Xt(),e,n)}if(r==="webnn"){if(typeof navigator>"u"||!navigator.ml)throw new Error("WebNN is not supported in current environment");await o("webnn",Xt(),e)}}},Zs=new Map,rw=e=>{let r=Xt(),t=r.stackSave();try{let s=r.PTR_SIZE,o=r.stackAlloc(2*s);r._OrtGetInputOutputCount(e,o,o+s)!==0&&Wt("Can't get session input/output count.");let n=s===4?"i32":"i64";return[Number(r.getValue(o,n)),Number(r.getValue(o+s,n))]}finally{r.stackRestore(t)}},uu=(e,r)=>{let t=Xt(),s=t.stackSave(),o=0;try{let n=t.PTR_SIZE,a=t.stackAlloc(2*n);t._OrtGetInputOutputMetadata(e,r,a,a+n)!==0&&Wt("Can't get session input/output metadata.");let i=Number(t.getValue(a,"*"));o=Number(t.getValue(a+n,"*"));let l=t.HEAP32[o/4];if(l===0)return[i,0];let c=t.HEAPU32[o/4+1],p=[];for(let d=0;d<c;d++){let u=Number(t.getValue(o+8+d*n,"*"));p.push(u!==0?t.UTF8ToString(u):Number(t.getValue(o+8+(d+c)*n,"*")))}return[i,l,p]}finally{t.stackRestore(s),o!==0&&t._OrtFree(o)}},va=e=>{let r=Xt(),t=r._malloc(e.byteLength);if(t===0)throw new Error(`Can't create a session. failed to allocate a buffer of size ${e.byteLength}.`);return r.HEAPU8.set(e,t),[t,e.byteLength]},cu=async(e,r)=>{var d,u,_,f;let t,s,o=Xt();Array.isArray(e)?[t,s]=e:e.buffer===o.HEAPU8.buffer?[t,s]=[e.byteOffset,e.byteLength]:[t,s]=va(e);let n=0,a=0,i=0,l=[],c=[],p=[];try{if([a,l]=await Vd(r),(r==null?void 0:r.externalData)&&o.mountExternalData){let P=[];for(let F of r.externalData){let D=typeof F=="string"?F:F.path;P.push(el(typeof F=="string"?F:F.data).then(K=>{o.mountExternalData(D,K)}))}await Promise.all(P)}for(let P of(r==null?void 0:r.executionProviders)??[])if((typeof P=="string"?P:P.name)==="webnn"){if(o.shouldTransferToMLTensor=!1,typeof P!="string"){let F=P,D=F==null?void 0:F.context,K=F==null?void 0:F.gpuDevice,U=F==null?void 0:F.deviceType,j=F==null?void 0:F.powerPreference;D?o.currentContext=D:K?o.currentContext=await o.webnnCreateMLContext(K):o.currentContext=await o.webnnCreateMLContext({deviceType:U,powerPreference:j})}else o.currentContext=await o.webnnCreateMLContext();break}n=await o._OrtCreateSession(t,s,a),(d=o.webgpuOnCreateSession)==null||d.call(o,n),n===0&&Wt("Can't create a session."),(u=o.jsepOnCreateSession)==null||u.call(o),o.currentContext&&(o.webnnRegisterMLContext(n,o.currentContext),o.currentContext=void 0,o.shouldTransferToMLTensor=!0);let[b,A]=rw(n),g=!!(r!=null&&r.enableGraphCapture),y=[],C=[],x=[],M=[],T=[];for(let P=0;P<b;P++){let[F,D,K]=uu(n,P);F===0&&Wt("Can't get an input name."),c.push(F);let U=o.UTF8ToString(F);y.push(U),x.push(D===0?{name:U,isTensor:!1}:{name:U,isTensor:!0,type:Ns(D),shape:K})}for(let P=0;P<A;P++){let[F,D,K]=uu(n,P+b);F===0&&Wt("Can't get an output name."),p.push(F);let U=o.UTF8ToString(F);C.push(U),M.push(D===0?{name:U,isTensor:!1}:{name:U,isTensor:!0,type:Ns(D),shape:K});{if(g&&(r==null?void 0:r.preferredOutputLocation)===void 0){T.push("gpu-buffer");continue}let j=typeof(r==null?void 0:r.preferredOutputLocation)=="string"?r.preferredOutputLocation:((_=r==null?void 0:r.preferredOutputLocation)==null?void 0:_[U])??"cpu";if(j!=="cpu"&&j!=="cpu-pinned"&&j!=="gpu-buffer"&&j!=="ml-tensor")throw new Error(`Not supported preferred output location: ${j}.`);if(g&&j!=="gpu-buffer")throw new Error(`Not supported preferred output location: ${j}. Only 'gpu-buffer' location is supported when enableGraphCapture is true.`);T.push(j)}}let v=null;return T.some(P=>P==="gpu-buffer"||P==="ml-tensor")&&(i=o._OrtCreateBinding(n),i===0&&Wt("Can't create IO binding."),v={handle:i,outputPreferredLocations:T,outputPreferredLocationsEncoded:T.map(P=>Zi(P))}),Zs.set(n,[n,c,p,v,g,!1]),[n,y,C,x,M]}catch(b){throw c.forEach(A=>o._OrtFree(A)),p.forEach(A=>o._OrtFree(A)),i!==0&&o._OrtReleaseBinding(i)!==0&&Wt("Can't release IO binding."),n!==0&&o._OrtReleaseSession(n)!==0&&Wt("Can't release session."),b}finally{o._free(t),a!==0&&o._OrtReleaseSessionOptions(a)!==0&&Wt("Can't release session options."),l.forEach(b=>o._free(b)),(f=o.unmountExternalData)==null||f.call(o)}},du=e=>{var l,c,p;let r=Xt(),t=Zs.get(e);if(!t)throw new Error(`cannot release session. invalid session id: ${e}`);let[s,o,n,a,i]=t;a&&(i&&r._OrtClearBoundOutputs(a.handle)!==0&&Wt("Can't clear bound outputs."),r._OrtReleaseBinding(a.handle)!==0&&Wt("Can't release IO binding.")),(l=r.jsepOnReleaseSession)==null||l.call(r,e),(c=r.webnnOnReleaseSession)==null||c.call(r,e),(p=r.webgpuOnReleaseSession)==null||p.call(r,e),o.forEach(d=>r._OrtFree(d)),n.forEach(d=>r._OrtFree(d)),r._OrtReleaseSession(s)!==0&&Wt("Can't release session."),Zs.delete(e)},pu=async(e,r,t,s,o,n,a=!1)=>{if(!e){r.push(0);return}let i=Xt(),l=i.PTR_SIZE,c=e[0],p=e[1],d=e[3],u=d,_,f;if(c==="string"&&(d==="gpu-buffer"||d==="ml-tensor"))throw new Error("String tensor is not supported on GPU.");if(a&&d!=="gpu-buffer")throw new Error(`External buffer must be provided for input/output index ${n} when enableGraphCapture is true.`);if(d==="gpu-buffer"){let g=e[2].gpuBuffer;f=fn(Wn(c),p);{let y=i.jsepRegisterBuffer;if(!y)throw new Error('Tensor location "gpu-buffer" is not supported without using WebGPU.');_=y(s,n,g,f)}}else if(d==="ml-tensor"){let g=e[2].mlTensor;f=fn(Wn(c),p);let y=i.webnnRegisterMLTensor;if(!y)throw new Error('Tensor location "ml-tensor" is not supported without using WebNN.');_=y(s,g,Wn(c),p)}else{let g=e[2];if(Array.isArray(g)){f=l*g.length,_=i._malloc(f),t.push(_);for(let y=0;y<g.length;y++){if(typeof g[y]!="string")throw new TypeError(`tensor data at index ${y} is not a string`);i.setValue(_+y*l,ys(g[y],t),"*")}}else{let y=i.webnnIsGraphInput;if(c!=="string"&&y){let C=i.UTF8ToString(o);if(y(s,C)){let x=Wn(c);f=fn(x,p),u="ml-tensor";let M=i.webnnCreateTemporaryTensor,T=i.webnnUploadTensor;if(!M||!T)throw new Error('Tensor location "ml-tensor" is not supported without using WebNN.');let v=await M(s,x,p);T(v,new Uint8Array(g.buffer,g.byteOffset,g.byteLength)),_=v}else f=g.byteLength,_=i._malloc(f),t.push(_),i.HEAPU8.set(new Uint8Array(g.buffer,g.byteOffset,f),_)}else f=g.byteLength,_=i._malloc(f),t.push(_),i.HEAPU8.set(new Uint8Array(g.buffer,g.byteOffset,f),_)}}let b=i.stackSave(),A=i.stackAlloc(4*p.length);try{p.forEach((y,C)=>i.setValue(A+C*l,y,l===4?"i32":"i64"));let g=i._OrtCreateTensor(Wn(c),_,f,A,p.length,Zi(u));g===0&&Wt(`Can't create tensor for input/output. session=${s}, index=${n}.`),r.push(g)}finally{i.stackRestore(b)}},hu=async(e,r,t,s,o,n)=>{var K,U,j,ne;let a=Xt(),i=a.PTR_SIZE,l=Zs.get(e);if(!l)throw new Error(`cannot run inference. invalid session id: ${e}`);let c=l[0],p=l[1],d=l[2],u=l[3],_=l[4],f=l[5],b=r.length,A=s.length,g=0,y=[],C=[],x=[],M=[],T=a.stackSave(),v=a.stackAlloc(b*i),P=a.stackAlloc(b*i),F=a.stackAlloc(A*i),D=a.stackAlloc(A*i);try{[g,y]=zd(n);for(let Z=0;Z<b;Z++)await pu(t[Z],C,M,e,p[r[Z]],r[Z],_);for(let Z=0;Z<A;Z++)await pu(o[Z],x,M,e,d[s[Z]],b+s[Z],_);for(let Z=0;Z<b;Z++)a.setValue(v+Z*i,C[Z],"*"),a.setValue(P+Z*i,p[r[Z]],"*");for(let Z=0;Z<A;Z++)a.setValue(F+Z*i,x[Z],"*"),a.setValue(D+Z*i,d[s[Z]],"*");if(u&&!f){let{handle:Z,outputPreferredLocations:ae,outputPreferredLocationsEncoded:he}=u;if(p.length!==b)throw new Error(`input count from feeds (${b}) is expected to be always equal to model's input count (${p.length}).`);for(let Q=0;Q<b;Q++){let B=r[Q];await a._OrtBindInput(Z,p[B],C[Q])!==0&&Wt(`Can't bind input[${Q}] for session=${e}.`)}for(let Q=0;Q<A;Q++){let B=s[Q];(K=o[Q])!=null&&K[3]?a._OrtBindOutput(Z,d[B],x[Q],0)!==0&&Wt(`Can't bind pre-allocated output[${Q}] for session=${e}.`):a._OrtBindOutput(Z,d[B],0,he[B])!==0&&Wt(`Can't bind output[${Q}] to ${ae[Q]} for session=${e}.`)}Zs.set(e,[c,p,d,u,_,!0])}(U=a.jsepOnRunStart)==null||U.call(a,c),(j=a.webnnOnRunStart)==null||j.call(a,c);let q;u?q=await a._OrtRunWithBinding(c,u.handle,A,F,g):q=await a._OrtRun(c,P,v,b,D,A,F,g),q!==0&&Wt("failed to call OrtRun().");let te=[];for(let Z=0;Z<A;Z++){let ae=Number(a.getValue(F+Z*i,"*"));if(ae===x[Z]){te.push(o[Z]);continue}let he=a.stackSave(),Q=a.stackAlloc(4*i),B=!1,O,W=0;try{a._OrtGetTensorData(ae,Q,Q+i,Q+2*i,Q+3*i)!==0&&Wt(`Can't access output tensor data on index ${Z}.`);let N=i===4?"i32":"i64",J=Number(a.getValue(Q,N));W=a.getValue(Q+i,"*");let ie=a.getValue(Q+i*2,"*"),me=Number(a.getValue(Q+i*3,N)),Ae=[];for(let X=0;X<me;X++)Ae.push(Number(a.getValue(ie+X*i,N)));a._OrtFree(ie)!==0&&Wt("Can't free memory for tensor dims.");let Ve=Ae.reduce((X,z)=>X*z,1);O=Ns(J);let $e=u==null?void 0:u.outputPreferredLocations[s[Z]];if(O==="string"){if($e==="gpu-buffer"||$e==="ml-tensor")throw new Error("String tensor is not supported on GPU.");let X=[];for(let z=0;z<Ve;z++){let _e=a.getValue(W+z*i,"*"),Ee=a.getValue(W+(z+1)*i,"*"),Me=z===Ve-1?void 0:Ee-_e;X.push(a.UTF8ToString(_e,Me))}te.push([O,Ae,X,"cpu"])}else if($e==="gpu-buffer"&&Ve>0){let X=a.jsepGetBuffer;if(!X)throw new Error('preferredLocation "gpu-buffer" is not supported without using WebGPU.');let z=X(W),_e=fn(J,Ve);if(_e===void 0||!Ji(O))throw new Error(`Unsupported data type: ${O}`);B=!0,te.push([O,Ae,{gpuBuffer:z,download:a.jsepCreateDownloader(z,_e,O),dispose:()=>{a._OrtReleaseTensor(ae)!==0&&Wt("Can't release tensor.")}},"gpu-buffer"])}else if($e==="ml-tensor"&&Ve>0){let X=a.webnnEnsureTensor,z=a.webnnIsInt64Supported;if(!X||!z)throw new Error('preferredLocation "ml-tensor" is not supported without using WebNN.');if(fn(J,Ve)===void 0||!Yi(O))throw new Error(`Unsupported data type: ${O}`);if(O==="int64"&&!z(e))throw new Error('preferredLocation "ml-tensor" for int64 output is not supported by current WebNN Context.');let _e=await X(e,W,J,Ae,!1);B=!0,te.push([O,Ae,{mlTensor:_e,download:a.webnnCreateMLTensorDownloader(W,O),dispose:()=>{a.webnnReleaseTensorId(W),a._OrtReleaseTensor(ae)}},"ml-tensor"])}else{let X=Xi(O),z=new X(Ve);new Uint8Array(z.buffer,z.byteOffset,z.byteLength).set(a.HEAPU8.subarray(W,W+z.byteLength)),te.push([O,Ae,z,"cpu"])}}finally{a.stackRestore(he),O==="string"&&W&&a._free(W),B||a._OrtReleaseTensor(ae),(ne=a.webnnOnRunEnd)==null||ne.call(a,c)}}return u&&!_&&(a._OrtClearBoundOutputs(u.handle)!==0&&Wt("Can't clear bound outputs."),Zs.set(e,[c,p,d,u,_,!1])),te}finally{a.stackRestore(T),C.forEach(q=>a._OrtReleaseTensor(q)),x.forEach(q=>a._OrtReleaseTensor(q)),M.forEach(q=>a._free(q)),g!==0&&a._OrtReleaseRunOptions(g),y.forEach(q=>a._free(q))}},mu=e=>{let r=Xt(),t=Zs.get(e);if(!t)throw new Error("invalid session id");let s=t[0],o=r._OrtEndProfiling(s);o===0&&Wt("Can't get an profile file name."),r._OrtFree(o)},fu=e=>{let r=[];for(let t of e){let s=t[2];!Array.isArray(s)&&"buffer"in s&&r.push(s.buffer)}return r}}),en,Jr,qn,Po,Co,xa,_u,Ta,vn,xn,nw,ow,aw,iw,lw,uw,cw,dw,pw=Ue(()=>{bs(),sw(),mn(),Gi(),en=()=>!!Jt.wasm.proxy&&typeof document<"u",qn=!1,Po=!1,Co=!1,Ta=new Map,vn=(e,r)=>{let t=Ta.get(e);t?t.push(r):Ta.set(e,[r])},xn=()=>{if(qn||!Po||Co||!Jr)throw new Error("worker not ready")},nw=e=>{switch(e.data.type){case"init-wasm":qn=!1,e.data.err?(Co=!0,_u[1](e.data.err)):(Po=!0,_u[0]()),xa&&(URL.revokeObjectURL(xa),xa=void 0);break;case"init-ep":case"copy-from":case"create":case"release":case"run":case"end-profiling":{let r=Ta.get(e.data.type);e.data.err?r.shift()[1](e.data.err):r.shift()[0](e.data.out);break}}},ow=async()=>{if(!Po){if(qn)throw new Error("multiple calls to 'initWasm()' detected.");if(Co)throw new Error("previous call to 'initWasm()' failed.");if(qn=!0,en())return new Promise((e,r)=>{Jr==null||Jr.terminate(),Id().then(([t,s])=>{try{Jr=s,Jr.onerror=n=>r(n),Jr.onmessage=nw,_u=[e,r];let o={type:"init-wasm",in:Jt};!o.in.wasm.wasmPaths&&(t||ji)&&(o.in.wasm.wasmPaths={wasm:new URL("/assets/ort-wasm-simd-threaded.jsep-B0T3yYHD.wasm",self.location.href).href}),Jr.postMessage(o),xa=t}catch(o){r(o)}},r)});try{await qi(Jt.wasm),await iu(Jt),Po=!0}catch(e){throw Co=!0,e}finally{qn=!1}}},aw=async e=>{if(en())return xn(),new Promise((r,t)=>{vn("init-ep",[r,t]);let s={type:"init-ep",in:{epName:e,env:Jt}};Jr.postMessage(s)});await lu(Jt,e)},iw=async e=>en()?(xn(),new Promise((r,t)=>{vn("copy-from",[r,t]);let s={type:"copy-from",in:{buffer:e}};Jr.postMessage(s,[e.buffer])})):va(e),lw=async(e,r)=>{if(en()){if(r!=null&&r.preferredOutputLocation)throw new Error('session option "preferredOutputLocation" is not supported for proxy.');return xn(),new Promise((t,s)=>{vn("create",[t,s]);let o={type:"create",in:{model:e,options:{...r}}},n=[];e instanceof Uint8Array&&n.push(e.buffer),Jr.postMessage(o,n)})}else return cu(e,r)},uw=async e=>{if(en())return xn(),new Promise((r,t)=>{vn("release",[r,t]);let s={type:"release",in:e};Jr.postMessage(s)});du(e)},cw=async(e,r,t,s,o,n)=>{if(en()){if(t.some(a=>a[3]!=="cpu"))throw new Error("input tensor on GPU is not supported for proxy.");if(o.some(a=>a))throw new Error("pre-allocated output tensor is not supported for proxy.");return xn(),new Promise((a,i)=>{vn("run",[a,i]);let l=t,c={type:"run",in:{sessionId:e,inputIndices:r,inputs:l,outputIndices:s,options:n}};Jr.postMessage(c,fu(l))})}else return hu(e,r,t,s,o,n)},dw=async e=>{if(en())return xn(),new Promise((r,t)=>{vn("end-profiling",[r,t]);let s={type:"end-profiling",in:e};Jr.postMessage(s)});mu(e)}}),gu,hw,mw,BT=Ue(()=>{bs(),pw(),Mt(),Di(),Ud(),gu=(e,r)=>{switch(e.location){case"cpu":return[e.type,e.dims,e.data,"cpu"];case"gpu-buffer":return[e.type,e.dims,{gpuBuffer:e.gpuBuffer},"gpu-buffer"];case"ml-tensor":return[e.type,e.dims,{mlTensor:e.mlTensor},"ml-tensor"];default:throw new Error(`invalid data location: ${e.location} for ${r()}`)}},hw=e=>{switch(e[3]){case"cpu":return new ws(e[0],e[2],e[1]);case"gpu-buffer":{let r=e[0];if(!Ji(r))throw new Error(`not supported data type: ${r} for deserializing GPU tensor`);let{gpuBuffer:t,download:s,dispose:o}=e[2];return ws.fromGpuBuffer(t,{dataType:r,dims:e[1],download:s,dispose:o})}case"ml-tensor":{let r=e[0];if(!Yi(r))throw new Error(`not supported data type: ${r} for deserializing MLTensor tensor`);let{mlTensor:t,download:s,dispose:o}=e[2];return ws.fromMLTensor(t,{dataType:r,dims:e[1],download:s,dispose:o})}default:throw new Error(`invalid data location: ${e[3]}`)}},mw=class{async fetchModelAndCopyToWasmMemory(e){return iw(await el(e))}async loadModel(e,r){Ms();let t;typeof e=="string"?t=await this.fetchModelAndCopyToWasmMemory(e):t=e,[this.sessionId,this.inputNames,this.outputNames,this.inputMetadata,this.outputMetadata]=await lw(t,r),us()}async dispose(){return uw(this.sessionId)}async run(e,r,t){Ms();let s=[],o=[];Object.entries(e).forEach(d=>{let u=d[0],_=d[1],f=this.inputNames.indexOf(u);if(f===-1)throw new Error(`invalid input '${u}'`);s.push(_),o.push(f)});let n=[],a=[];Object.entries(r).forEach(d=>{let u=d[0],_=d[1],f=this.outputNames.indexOf(u);if(f===-1)throw new Error(`invalid output '${u}'`);n.push(_),a.push(f)});let i=s.map((d,u)=>gu(d,()=>`input "${this.inputNames[o[u]]}"`)),l=n.map((d,u)=>d?gu(d,()=>`output "${this.outputNames[a[u]]}"`):null),c=await cw(this.sessionId,o,i,a,l,t),p={};for(let d=0;d<c.length;d++)p[this.outputNames[a[d]]]=n[d]??hw(c[d]);return us(),p}startProfiling(){}endProfiling(){dw(this.sessionId)}}}),fw={};Un(fw,{OnnxruntimeWebAssemblyBackend:()=>Mu,initializeFlags:()=>wu,wasmBackend:()=>_w});var wu,Mu,_w,NT=Ue(()=>{bs(),pw(),BT(),wu=()=>{(typeof Jt.wasm.initTimeout!="number"||Jt.wasm.initTimeout<0)&&(Jt.wasm.initTimeout=0);let e=Jt.wasm.simd;if(typeof e!="boolean"&&e!==void 0&&e!=="fixed"&&e!=="relaxed"&&(console.warn(`Property "env.wasm.simd" is set to unknown value "${e}". Reset it to \`false\` and ignore SIMD feature checking.`),Jt.wasm.simd=!1),typeof Jt.wasm.proxy!="boolean"&&(Jt.wasm.proxy=!1),typeof Jt.wasm.trace!="boolean"&&(Jt.wasm.trace=!1),typeof Jt.wasm.numThreads!="number"||!Number.isInteger(Jt.wasm.numThreads)||Jt.wasm.numThreads<=0)if(typeof self<"u"&&!self.crossOriginIsolated)Jt.wasm.numThreads=1;else{let r=typeof navigator>"u"?Px("node:os").cpus().length:navigator.hardwareConcurrency;Jt.wasm.numThreads=Math.min(4,Math.ceil((r||1)/2))}},Mu=class{async init(e){wu(),await ow(),await aw(e)}async createInferenceSessionHandler(e,r){let t=new mw;return await t.loadModel(e,r),t}},_w=new Mu});bs(),bs(),bs();var jT="1.22.0-dev.20250409-89f8206ba4",VT=yd;{let e=(NT(),co(fw)).wasmBackend;pn("webgpu",e,5),pn("webnn",e,5),pn("cpu",e,10),pn("wasm",e,10)}Object.defineProperty(Jt.versions,"web",{value:jT,enumerable:!0});/**
* @license
* Copyright 2021 Google LLC. All Rights Reserved.
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
* =============================================================================
*//**
 * @license
 * Copyright 2020 Google LLC. All Rights Reserved.
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 * http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 * =============================================================================
 *//**
 * @license
 * Copyright 2019 Google LLC. All Rights Reserved.
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 * http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 * =============================================================================
 */var UT=Object.freeze({__proto__:null,get InferenceSession(){return Oi},get TRACE(){return mo},get TRACE_FUNC_BEGIN(){return Ms},get TRACE_FUNC_END(){return us},get Tensor(){return ws},default:VT,get env(){return Jt},get registerBackend(){return pn}}),Us={},WT={"onnxruntime-common":e=>{e.exports=vx},"onnxruntime-web":e=>{e.exports=UT},"?2ce3":()=>{},"?7992":()=>{},"?5af5":()=>{},"?2b25":()=>{},"?db59":()=>{},"?383f":()=>{},"?fa4b":()=>{},"./node_modules/@huggingface/jinja/dist/index.js":(e,r,t)=>{t.r(r),t.d(r,{Environment:()=>ze,Interpreter:()=>gt,Template:()=>Kt,parse:()=>ie,tokenize:()=>p});var s=Object.freeze({Text:"Text",NumericLiteral:"NumericLiteral",StringLiteral:"StringLiteral",Identifier:"Identifier",Equals:"Equals",OpenParen:"OpenParen",CloseParen:"CloseParen",OpenStatement:"OpenStatement",CloseStatement:"CloseStatement",OpenExpression:"OpenExpression",CloseExpression:"CloseExpression",OpenSquareBracket:"OpenSquareBracket",CloseSquareBracket:"CloseSquareBracket",OpenCurlyBracket:"OpenCurlyBracket",CloseCurlyBracket:"CloseCurlyBracket",Comma:"Comma",Dot:"Dot",Colon:"Colon",Pipe:"Pipe",CallOperator:"CallOperator",AdditiveBinaryOperator:"AdditiveBinaryOperator",MultiplicativeBinaryOperator:"MultiplicativeBinaryOperator",ComparisonBinaryOperator:"ComparisonBinaryOperator",UnaryOperator:"UnaryOperator",Comment:"Comment"}),o=class{constructor($,ee){this.value=$,this.type=ee}};function n($){return/\w/.test($)}function a($){return/[0-9]/.test($)}var i=[["{%",s.OpenStatement],["%}",s.CloseStatement],["{{",s.OpenExpression],["}}",s.CloseExpression],["(",s.OpenParen],[")",s.CloseParen],["{",s.OpenCurlyBracket],["}",s.CloseCurlyBracket],["[",s.OpenSquareBracket],["]",s.CloseSquareBracket],[",",s.Comma],[".",s.Dot],[":",s.Colon],["|",s.Pipe],["<=",s.ComparisonBinaryOperator],[">=",s.ComparisonBinaryOperator],["==",s.ComparisonBinaryOperator],["!=",s.ComparisonBinaryOperator],["<",s.ComparisonBinaryOperator],[">",s.ComparisonBinaryOperator],["+",s.AdditiveBinaryOperator],["-",s.AdditiveBinaryOperator],["~",s.AdditiveBinaryOperator],["*",s.MultiplicativeBinaryOperator],["/",s.MultiplicativeBinaryOperator],["%",s.MultiplicativeBinaryOperator],["=",s.Equals]],l=new Map([["n",`
`],["t","	"],["r","\r"],["b","\b"],["f","\f"],["v","\v"],["'","'"],['"','"'],["\\","\\"]]);function c($,ee={}){return $.endsWith(`
`)&&($=$.slice(0,-1)),ee.lstrip_blocks&&($=$.replace(/^[ \t]*({[#%-])/gm,"$1")),ee.trim_blocks&&($=$.replace(/([#%-]})\n/g,"$1")),$.replace(/-%}\s*/g,"%}").replace(/\s*{%-/g,"{%").replace(/-}}\s*/g,"}}").replace(/\s*{{-/g,"{{").replace(/-#}\s*/g,"#}").replace(/\s*{#-/g,"{#").replace(/{%\s*(end)?generation\s*%}/gs,"")}function p($,ee={}){var nt,wt;const V=[],Y=c($,ee);let oe=0,xe=0;const De=pt=>{let xt="";for(;pt(Y[oe]);){if(Y[oe]==="\\"){if(++oe,oe>=Y.length)throw new SyntaxError("Unexpected end of input");const tt=Y[oe++],It=l.get(tt);if(It===void 0)throw new SyntaxError(`Unexpected escaped character: ${tt}`);xt+=It;continue}if(xt+=Y[oe++],oe>=Y.length)throw new SyntaxError("Unexpected end of input")}return xt};e:for(;oe<Y.length;){const pt=(nt=V.at(-1))==null?void 0:nt.type;if(pt===void 0||pt===s.CloseStatement||pt===s.CloseExpression||pt===s.Comment){let tt="";for(;oe<Y.length&&!(Y[oe]==="{"&&(Y[oe+1]==="%"||Y[oe+1]==="{"||Y[oe+1]==="#"));)tt+=Y[oe++];if(tt.length>0){V.push(new o(tt,s.Text));continue}}if(Y[oe]==="{"&&Y[oe+1]==="#"){oe+=2;let tt="";for(;Y[oe]!=="#"||Y[oe+1]!=="}";){if(oe+2>=Y.length)throw new SyntaxError("Missing end of comment tag");tt+=Y[oe++]}V.push(new o(tt,s.Comment)),oe+=2;continue}De(tt=>/\s/.test(tt));const xt=Y[oe];if(xt==="-"||xt==="+"){const tt=(wt=V.at(-1))==null?void 0:wt.type;if(tt===s.Text||tt===void 0)throw new SyntaxError(`Unexpected character: ${xt}`);switch(tt){case s.Identifier:case s.NumericLiteral:case s.StringLiteral:case s.CloseParen:case s.CloseSquareBracket:break;default:{++oe;const It=De(a);V.push(new o(`${xt}${It}`,It.length>0?s.NumericLiteral:s.UnaryOperator));continue}}}for(const[tt,It]of i){if(tt==="}}"&&xe>0)continue;if(Y.slice(oe,oe+tt.length)===tt){V.push(new o(tt,It)),It===s.OpenExpression?xe=0:It===s.OpenCurlyBracket?++xe:It===s.CloseCurlyBracket&&--xe,oe+=tt.length;continue e}}if(xt==="'"||xt==='"'){++oe;const tt=De(It=>It!==xt);V.push(new o(tt,s.StringLiteral)),++oe;continue}if(a(xt)){let tt=De(a);if(Y[oe]==="."&&a(Y[oe+1])){++oe;const It=De(a);tt=`${tt}.${It}`}V.push(new o(tt,s.NumericLiteral));continue}if(n(xt)){const tt=De(n);V.push(new o(tt,s.Identifier));continue}throw new SyntaxError(`Unexpected character: ${xt}`)}return V}var d=class{constructor(){re(this,"type","Statement")}},u=class extends d{constructor(ee){super();re(this,"type","Program");this.body=ee}},_=class extends d{constructor(ee,V,Y){super();re(this,"type","If");this.test=ee,this.body=V,this.alternate=Y}},f=class extends d{constructor(ee,V,Y,oe){super();re(this,"type","For");this.loopvar=ee,this.iterable=V,this.body=Y,this.defaultBlock=oe}},b=class extends d{constructor(){super(...arguments);re(this,"type","Break")}},A=class extends d{constructor(){super(...arguments);re(this,"type","Continue")}},g=class extends d{constructor(ee,V,Y){super();re(this,"type","Set");this.assignee=ee,this.value=V,this.body=Y}},y=class extends d{constructor(ee,V,Y){super();re(this,"type","Macro");this.name=ee,this.args=V,this.body=Y}},C=class extends d{constructor(ee){super();re(this,"type","Comment");this.value=ee}},x=class extends d{constructor(){super(...arguments);re(this,"type","Expression")}},M=class extends x{constructor(ee,V,Y){super();re(this,"type","MemberExpression");this.object=ee,this.property=V,this.computed=Y}},T=class extends x{constructor(ee,V){super();re(this,"type","CallExpression");this.callee=ee,this.args=V}},v=class extends x{constructor(ee){super();re(this,"type","Identifier");this.value=ee}},P=class extends x{constructor(ee){super();re(this,"type","Literal");this.value=ee}},F=class extends P{constructor(){super(...arguments);re(this,"type","IntegerLiteral")}},D=class extends P{constructor(){super(...arguments);re(this,"type","FloatLiteral")}},K=class extends P{constructor(){super(...arguments);re(this,"type","StringLiteral")}},U=class extends P{constructor(){super(...arguments);re(this,"type","ArrayLiteral")}},j=class extends P{constructor(){super(...arguments);re(this,"type","TupleLiteral")}},ne=class extends P{constructor(){super(...arguments);re(this,"type","ObjectLiteral")}},q=class extends x{constructor(ee,V,Y){super();re(this,"type","BinaryExpression");this.operator=ee,this.left=V,this.right=Y}},te=class extends x{constructor(ee,V){super();re(this,"type","FilterExpression");this.operand=ee,this.filter=V}},Z=class extends d{constructor(ee,V){super();re(this,"type","FilterStatement");this.filter=ee,this.body=V}},ae=class extends x{constructor(ee,V){super();re(this,"type","SelectExpression");this.lhs=ee,this.test=V}},he=class extends x{constructor(ee,V,Y){super();re(this,"type","TestExpression");this.operand=ee,this.negate=V,this.test=Y}},Q=class extends x{constructor(ee,V){super();re(this,"type","UnaryExpression");this.operator=ee,this.argument=V}},B=class extends x{constructor(ee=void 0,V=void 0,Y=void 0){super();re(this,"type","SliceExpression");this.start=ee,this.stop=V,this.step=Y}},O=class extends x{constructor(ee,V){super();re(this,"type","KeywordArgumentExpression");this.key=ee,this.value=V}},W=class extends x{constructor(ee){super();re(this,"type","SpreadExpression");this.argument=ee}},N=class extends d{constructor(ee,V,Y){super();re(this,"type","CallStatement");this.call=ee,this.callerArgs=V,this.body=Y}},J=class extends x{constructor(ee,V,Y){super();re(this,"type","Ternary");this.condition=ee,this.trueExpr=V,this.falseExpr=Y}};function ie($){const ee=new u([]);let V=0;function Y(Ne,je){const rt=$[V++];if(!rt||rt.type!==Ne)throw new Error(`Parser Error: ${je}. ${rt.type} !== ${Ne}.`);return rt}function oe(Ne){if(!wt(Ne))throw new SyntaxError(`Expected ${Ne}`);++V}function xe(){switch($[V].type){case s.Comment:return new C($[V++].value);case s.Text:return pt();case s.OpenStatement:return xt();case s.OpenExpression:return tt();default:throw new SyntaxError(`Unexpected token type: ${$[V].type}`)}}function De(...Ne){return V+Ne.length<=$.length&&Ne.every((je,rt)=>je===$[V+rt].type)}function nt(...Ne){var je,rt,Qt;return((je=$[V])==null?void 0:je.type)===s.OpenStatement&&((rt=$[V+1])==null?void 0:rt.type)===s.Identifier&&Ne.includes((Qt=$[V+1])==null?void 0:Qt.value)}function wt(...Ne){return V+Ne.length<=$.length&&Ne.every((je,rt)=>$[V+rt].type==="Identifier"&&je===$[V+rt].value)}function pt(){return new K(Y(s.Text,"Expected text token").value)}function xt(){if(Y(s.OpenStatement,"Expected opening statement token"),$[V].type!==s.Identifier)throw new SyntaxError(`Unknown statement, got ${$[V].type}`);const Ne=$[V].value;let je;switch(Ne){case"set":++V,je=It();break;case"if":++V,je=qt(),Y(s.OpenStatement,"Expected {% token"),oe("endif"),Y(s.CloseStatement,"Expected %} token");break;case"macro":++V,je=Wr(),Y(s.OpenStatement,"Expected {% token"),oe("endmacro"),Y(s.CloseStatement,"Expected %} token");break;case"for":++V,je=nr(),Y(s.OpenStatement,"Expected {% token"),oe("endfor"),Y(s.CloseStatement,"Expected %} token");break;case"call":{++V;let rt=null;De(s.OpenParen)&&(rt=Yr());const Qt=Gr();if(Qt.type!=="Identifier")throw new SyntaxError("Expected identifier following call statement");const Hs=Yr();Y(s.CloseStatement,"Expected closing statement token");const Ss=[];for(;!nt("endcall");)Ss.push(xe());Y(s.OpenStatement,"Expected '{%'"),oe("endcall"),Y(s.CloseStatement,"Expected closing statement token");const ss=new T(Qt,Hs);je=new N(ss,rt,Ss);break}case"break":++V,Y(s.CloseStatement,"Expected closing statement token"),je=new b;break;case"continue":++V,Y(s.CloseStatement,"Expected closing statement token"),je=new A;break;case"filter":{++V;let rt=Gr();rt instanceof v&&De(s.OpenParen)&&(rt=ms(rt)),Y(s.CloseStatement,"Expected closing statement token");const Qt=[];for(;!nt("endfilter");)Qt.push(xe());Y(s.OpenStatement,"Expected '{%'"),oe("endfilter"),Y(s.CloseStatement,"Expected '%}'"),je=new Z(rt,Qt);break}default:throw new SyntaxError(`Unknown statement type: ${Ne}`)}return je}function tt(){Y(s.OpenExpression,"Expected opening expression token");const Ne=kr();return Y(s.CloseExpression,"Expected closing expression token"),Ne}function It(){const Ne=qr();let je=null;const rt=[];if(De(s.Equals))++V,je=qr();else{for(Y(s.CloseStatement,"Expected %} token");!nt("endset");)rt.push(xe());Y(s.OpenStatement,"Expected {% token"),oe("endset")}return Y(s.CloseStatement,"Expected closing statement token"),new g(Ne,je,rt)}function qt(){const Ne=kr();Y(s.CloseStatement,"Expected closing statement token");const je=[],rt=[];for(;!nt("elif","else","endif");)je.push(xe());if(nt("elif")){++V,++V;const Qt=qt();rt.push(Qt)}else if(nt("else"))for(++V,++V,Y(s.CloseStatement,"Expected closing statement token");!nt("endif");)rt.push(xe());return new _(Ne,je,rt)}function Wr(){const Ne=Gr();if(Ne.type!=="Identifier")throw new SyntaxError("Expected identifier following macro statement");const je=Yr();Y(s.CloseStatement,"Expected closing statement token");const rt=[];for(;!nt("endmacro");)rt.push(xe());return new y(Ne,je,rt)}function qr(Ne=!1){const je=Ne?Gr:kr,rt=[je()],Qt=De(s.Comma);for(;Qt&&(++V,rt.push(je()),!!De(s.Comma)););return Qt?new j(rt):rt[0]}function nr(){const Ne=qr(!0);if(!(Ne instanceof v||Ne instanceof j))throw new SyntaxError(`Expected identifier/tuple for the loop variable, got ${Ne.type} instead`);if(!wt("in"))throw new SyntaxError("Expected `in` keyword following loop variable");++V;const je=kr();Y(s.CloseStatement,"Expected closing statement token");const rt=[];for(;!nt("endfor","else");)rt.push(xe());const Qt=[];if(nt("else"))for(++V,++V,Y(s.CloseStatement,"Expected closing statement token");!nt("endfor");)Qt.push(xe());return new f(Ne,je,rt,Qt)}function kr(){return cr()}function cr(){const Ne=ps();if(wt("if")){++V;const je=ps();if(wt("else")){++V;const rt=cr();return new J(je,Ne,rt)}else return new ae(Ne,je)}return Ne}function ps(){let Ne=hs();for(;wt("or");){const je=$[V];++V;const rt=hs();Ne=new q(je,Ne,rt)}return Ne}function hs(){let Ne=Ir();for(;wt("and");){const je=$[V];++V;const rt=Ir();Ne=new q(je,Ne,rt)}return Ne}function Ir(){let Ne;for(;wt("not");){const je=$[V];++V;const rt=Ir();Ne=new Q(je,rt)}return Ne??Ls()}function Ls(){let Ne=zs();for(;;){let je;if(wt("not","in"))je=new o("not in",s.Identifier),V+=2;else if(wt("in"))je=$[V++];else if(De(s.ComparisonBinaryOperator))je=$[V++];else break;const rt=zs();Ne=new q(je,Ne,rt)}return Ne}function zs(){let Ne=ar();for(;De(s.AdditiveBinaryOperator);){const je=$[V];++V;const rt=ar();Ne=new q(je,Ne,rt)}return Ne}function vr(){const Ne=mr(Gr());return De(s.OpenParen)?ms(Ne):Ne}function ms(Ne){let je=new T(Ne,Yr());return je=mr(je),De(s.OpenParen)&&(je=ms(je)),je}function Yr(){Y(s.OpenParen,"Expected opening parenthesis for arguments list");const Ne=Rr();return Y(s.CloseParen,"Expected closing parenthesis for arguments list"),Ne}function Rr(){const Ne=[];for(;!De(s.CloseParen);){let je;if($[V].type===s.MultiplicativeBinaryOperator&&$[V].value==="*"){++V;const rt=kr();je=new W(rt)}else if(je=kr(),De(s.Equals)){if(++V,!(je instanceof v))throw new SyntaxError("Expected identifier for keyword argument");const rt=kr();je=new O(je,rt)}Ne.push(je),De(s.Comma)&&++V}return Ne}function Cs(){const Ne=[];let je=!1;for(;!De(s.CloseSquareBracket);)De(s.Colon)?(Ne.push(void 0),++V,je=!0):(Ne.push(kr()),De(s.Colon)&&(++V,je=!0));if(Ne.length===0)throw new SyntaxError("Expected at least one argument for member/slice expression");if(je){if(Ne.length>3)throw new SyntaxError("Expected 0-3 arguments for slice expression");return new B(...Ne)}return Ne[0]}function mr(Ne){for(;De(s.Dot)||De(s.OpenSquareBracket);){const je=$[V];++V;let rt;const Qt=je.type===s.OpenSquareBracket;if(Qt)rt=Cs(),Y(s.CloseSquareBracket,"Expected closing square bracket");else if(rt=Gr(),rt.type!=="Identifier")throw new SyntaxError("Expected identifier following dot operator");Ne=new M(Ne,rt,Qt)}return Ne}function ar(){let Ne=fs();for(;De(s.MultiplicativeBinaryOperator);){const je=$[V++],rt=fs();Ne=new q(je,Ne,rt)}return Ne}function fs(){let Ne=Gs();for(;wt("is");){++V;const je=wt("not");je&&++V;const rt=Gr();if(!(rt instanceof v))throw new SyntaxError("Expected identifier for the test");Ne=new he(Ne,je,rt)}return Ne}function Gs(){let Ne=vr();for(;De(s.Pipe);){++V;let je=Gr();if(!(je instanceof v))throw new SyntaxError("Expected identifier for the filter");De(s.OpenParen)&&(je=ms(je)),Ne=new te(Ne,je)}return Ne}function Gr(){const Ne=$[V++];switch(Ne.type){case s.NumericLiteral:{const je=Ne.value;return je.includes(".")?new D(Number(je)):new F(Number(je))}case s.StringLiteral:{let je=Ne.value;for(;De(s.StringLiteral);)je+=$[V++].value;return new K(je)}case s.Identifier:return new v(Ne.value);case s.OpenParen:{const je=qr();return Y(s.CloseParen,"Expected closing parenthesis, got ${tokens[current].type} instead."),je}case s.OpenSquareBracket:{const je=[];for(;!De(s.CloseSquareBracket);)je.push(kr()),De(s.Comma)&&++V;return++V,new U(je)}case s.OpenCurlyBracket:{const je=new Map;for(;!De(s.CloseCurlyBracket);){const rt=kr();Y(s.Colon,"Expected colon between key and value in object literal");const Qt=kr();je.set(rt,Qt),De(s.Comma)&&++V}return++V,new ne(je)}default:throw new SyntaxError(`Unexpected token: ${Ne.type}`)}}for(;V<$.length;)ee.body.push(xe());return ee}function me($,ee,V=1){ee===void 0&&(ee=$,$=0);const Y=[];for(let oe=$;oe<ee;oe+=V)Y.push(oe);return Y}function Ae($,ee,V,Y=1){const oe=Math.sign(Y);oe>=0?(ee=(ee??(ee=0))<0?Math.max($.length+ee,0):Math.min(ee,$.length),V=(V??(V=$.length))<0?Math.max($.length+V,0):Math.min(V,$.length)):(ee=(ee??(ee=$.length-1))<0?Math.max($.length+ee,-1):Math.min(ee,$.length-1),V=(V??(V=-1))<-1?Math.max($.length+V,-1):Math.min(V,$.length-1));const xe=[];for(let De=ee;oe*De<oe*V;De+=Y)xe.push($[De]);return xe}function Ve($){return $.replace(/\b\w/g,ee=>ee.toUpperCase())}function $e($){return X(new Date,$)}function X($,ee){const V=new Intl.DateTimeFormat(void 0,{month:"long"}),Y=new Intl.DateTimeFormat(void 0,{month:"short"}),oe=xe=>xe<10?"0"+xe:xe.toString();return ee.replace(/%[YmdbBHM%]/g,xe=>{switch(xe){case"%Y":return $.getFullYear().toString();case"%m":return oe($.getMonth()+1);case"%d":return oe($.getDate());case"%b":return Y.format($);case"%B":return V.format($);case"%H":return oe($.getHours());case"%M":return oe($.getMinutes());case"%%":return"%";default:return xe}})}function z($){return $.replace(/[.*+?^${}()|[\]\\]/g,"\\$&")}function _e($,ee,V,Y){if(Y===0)return $;let oe=Y==null||Y<0?1/0:Y;const xe=ee.length===0?new RegExp("(?=)","gu"):new RegExp(z(ee),"gu");return $.replaceAll(xe,De=>oe>0?(--oe,V):De)}var Ee=class extends Error{},Me=class extends Error{},Ce=class{constructor($=void 0){re(this,"type","RuntimeValue");re(this,"value");re(this,"builtins",new Map);this.value=$}__bool__(){return new ce(!!this.value)}toString(){return String(this.value)}},ye=class extends Ce{constructor(){super(...arguments);re(this,"type","IntegerValue")}},de=class extends Ce{constructor(){super(...arguments);re(this,"type","FloatValue")}toString(){return this.value%1===0?this.value.toFixed(1):this.value.toString()}},we=class extends Ce{constructor(){super(...arguments);re(this,"type","StringValue");re(this,"builtins",new Map([["upper",new qe(()=>new we(this.value.toUpperCase()))],["lower",new qe(()=>new we(this.value.toLowerCase()))],["strip",new qe(()=>new we(this.value.trim()))],["title",new qe(()=>new we(Ve(this.value)))],["capitalize",new qe(()=>new we(this.value.charAt(0).toUpperCase()+this.value.slice(1)))],["length",new ye(this.value.length)],["rstrip",new qe(()=>new we(this.value.trimEnd()))],["lstrip",new qe(()=>new we(this.value.trimStart()))],["startswith",new qe(ee=>{if(ee.length===0)throw new Error("startswith() requires at least one argument");const V=ee[0];if(V instanceof we)return new ce(this.value.startsWith(V.value));if(V instanceof Te){for(const Y of V.value){if(!(Y instanceof we))throw new Error("startswith() tuple elements must be strings");if(this.value.startsWith(Y.value))return new ce(!0)}return new ce(!1)}throw new Error("startswith() argument must be a string or tuple of strings")})],["endswith",new qe(ee=>{if(ee.length===0)throw new Error("endswith() requires at least one argument");const V=ee[0];if(V instanceof we)return new ce(this.value.endsWith(V.value));if(V instanceof Te){for(const Y of V.value){if(!(Y instanceof we))throw new Error("endswith() tuple elements must be strings");if(this.value.endsWith(Y.value))return new ce(!0)}return new ce(!1)}throw new Error("endswith() argument must be a string or tuple of strings")})],["split",new qe(ee=>{const V=ee[0]??new st;if(!(V instanceof we||V instanceof st))throw new Error("sep argument must be a string or null");const Y=ee[1]??new ye(-1);if(!(Y instanceof ye))throw new Error("maxsplit argument must be a number");let oe=[];if(V instanceof st){const xe=this.value.trimStart();for(const{0:De,index:nt}of xe.matchAll(/\S+/g)){if(Y.value!==-1&&oe.length>=Y.value&&nt!==void 0){oe.push(De+xe.slice(nt+De.length));break}oe.push(De)}}else{if(V.value==="")throw new Error("empty separator");oe=this.value.split(V.value),Y.value!==-1&&oe.length>Y.value&&oe.push(oe.splice(Y.value).join(V.value))}return new Te(oe.map(xe=>new we(xe)))})],["replace",new qe(ee=>{if(ee.length<2)throw new Error("replace() requires at least two arguments");const V=ee[0],Y=ee[1];if(!(V instanceof we&&Y instanceof we))throw new Error("replace() arguments must be strings");let oe;if(ee.length>2?ee[2].type==="KeywordArgumentsValue"?oe=ee[2].value.get("count")??new st:oe=ee[2]:oe=new st,!(oe instanceof ye||oe instanceof st))throw new Error("replace() count argument must be a number or null");return new we(_e(this.value,V.value,Y.value,oe.value))})]]))}},ce=class extends Ce{constructor(){super(...arguments);re(this,"type","BooleanValue")}},ke=class extends Ce{constructor(){super(...arguments);re(this,"type","ObjectValue");re(this,"builtins",new Map([["get",new qe(([ee,V])=>{if(!(ee instanceof we))throw new Error(`Object key must be a string: got ${ee.type}`);return this.value.get(ee.value)??V??new st})],["items",new qe(()=>this.items())],["keys",new qe(()=>this.keys())],["values",new qe(()=>this.values())]]))}__bool__(){return new ce(this.value.size>0)}items(){return new Te(Array.from(this.value.entries()).map(([ee,V])=>new Te([new we(ee),V])))}keys(){return new Te(Array.from(this.value.keys()).map(ee=>new we(ee)))}values(){return new Te(Array.from(this.value.values()))}},Le=class extends ke{constructor(){super(...arguments);re(this,"type","KeywordArgumentsValue")}},Te=class extends Ce{constructor(){super(...arguments);re(this,"type","ArrayValue");re(this,"builtins",new Map([["length",new ye(this.value.length)]]))}__bool__(){return new ce(this.value.length>0)}},We=class extends Te{constructor(){super(...arguments);re(this,"type","TupleValue")}},qe=class extends Ce{constructor(){super(...arguments);re(this,"type","FunctionValue")}},st=class extends Ce{constructor(){super(...arguments);re(this,"type","NullValue")}},Ze=class extends Ce{constructor(){super(...arguments);re(this,"type","UndefinedValue")}},ze=class{constructor($){re(this,"variables",new Map([["namespace",new qe($=>{if($.length===0)return new ke(new Map);if($.length!==1||!($[0]instanceof ke))throw new Error("`namespace` expects either zero arguments or a single object argument");return $[0]})]]));re(this,"tests",new Map([["boolean",$=>$.type==="BooleanValue"],["callable",$=>$ instanceof qe],["odd",$=>{if(!($ instanceof ye))throw new Error(`cannot odd on ${$.type}`);return $.value%2!==0}],["even",$=>{if(!($ instanceof ye))throw new Error(`cannot even on ${$.type}`);return $.value%2===0}],["false",$=>$.type==="BooleanValue"&&!$.value],["true",$=>$.type==="BooleanValue"&&$.value],["none",$=>$.type==="NullValue"],["string",$=>$.type==="StringValue"],["number",$=>$ instanceof ye||$ instanceof de],["integer",$=>$ instanceof ye],["iterable",$=>$.type==="ArrayValue"||$.type==="StringValue"],["mapping",$=>$.type==="ObjectValue"],["lower",$=>{const ee=$.value;return $.type==="StringValue"&&ee===ee.toLowerCase()}],["upper",$=>{const ee=$.value;return $.type==="StringValue"&&ee===ee.toUpperCase()}],["none",$=>$.type==="NullValue"],["defined",$=>$.type!=="UndefinedValue"],["undefined",$=>$.type==="UndefinedValue"],["equalto",($,ee)=>$.value===ee.value],["eq",($,ee)=>$.value===ee.value]]));this.parent=$}set($,ee){return this.declareVariable($,dt(ee))}declareVariable($,ee){if(this.variables.has($))throw new SyntaxError(`Variable already declared: ${$}`);return this.variables.set($,ee),ee}setVariable($,ee){return this.variables.set($,ee),ee}resolve($){if(this.variables.has($))return this;if(this.parent)return this.parent.resolve($);throw new Error(`Unknown variable: ${$}`)}lookupVariable($){try{return this.resolve($).variables.get($)??new Ze}catch{return new Ze}}};function He($){$.set("false",!1),$.set("true",!0),$.set("none",null),$.set("raise_exception",ee=>{throw new Error(ee)}),$.set("range",me),$.set("strftime_now",$e),$.set("True",!0),$.set("False",!1),$.set("None",null)}var gt=class{constructor($){re(this,"global");this.global=$??new ze}run($){return this.evaluate($,this.global)}evaluateBinaryExpression($,ee){const V=this.evaluate($.left,ee);switch($.operator.value){case"and":return V.__bool__().value?this.evaluate($.right,ee):V;case"or":return V.__bool__().value?V:this.evaluate($.right,ee)}const Y=this.evaluate($.right,ee);switch($.operator.value){case"==":return new ce(V.value==Y.value);case"!=":return new ce(V.value!=Y.value)}if(V instanceof Ze||Y instanceof Ze){if(Y instanceof Ze&&["in","not in"].includes($.operator.value))return new ce($.operator.value==="not in");throw new Error(`Cannot perform operation ${$.operator.value} on undefined values`)}else{if(V instanceof st||Y instanceof st)throw new Error("Cannot perform operation on null values");if($.operator.value==="~")return new we(V.value.toString()+Y.value.toString());if((V instanceof ye||V instanceof de)&&(Y instanceof ye||Y instanceof de)){const oe=V.value,xe=Y.value;switch($.operator.value){case"+":case"-":case"*":{const De=$.operator.value==="+"?oe+xe:$.operator.value==="-"?oe-xe:oe*xe;return V instanceof de||Y instanceof de?new de(De):new ye(De)}case"/":return new de(oe/xe);case"%":{const De=oe%xe;return V instanceof de||Y instanceof de?new de(De):new ye(De)}case"<":return new ce(oe<xe);case">":return new ce(oe>xe);case">=":return new ce(oe>=xe);case"<=":return new ce(oe<=xe)}}else if(V instanceof Te&&Y instanceof Te)switch($.operator.value){case"+":return new Te(V.value.concat(Y.value))}else if(Y instanceof Te){const oe=Y.value.find(xe=>xe.value===V.value)!==void 0;switch($.operator.value){case"in":return new ce(oe);case"not in":return new ce(!oe)}}}if(V instanceof we||Y instanceof we)switch($.operator.value){case"+":return new we(V.value.toString()+Y.value.toString())}if(V instanceof we&&Y instanceof we)switch($.operator.value){case"in":return new ce(Y.value.includes(V.value));case"not in":return new ce(!Y.value.includes(V.value))}if(V instanceof we&&Y instanceof ke)switch($.operator.value){case"in":return new ce(Y.value.has(V.value));case"not in":return new ce(!Y.value.has(V.value))}throw new SyntaxError(`Unknown operator "${$.operator.value}" between ${V.type} and ${Y.type}`)}evaluateArguments($,ee){const V=[],Y=new Map;for(const oe of $)if(oe.type==="SpreadExpression"){const xe=oe,De=this.evaluate(xe.argument,ee);if(!(De instanceof Te))throw new Error(`Cannot unpack non-iterable type: ${De.type}`);for(const nt of De.value)V.push(nt)}else if(oe.type==="KeywordArgumentExpression"){const xe=oe;Y.set(xe.key.value,this.evaluate(xe.value,ee))}else{if(Y.size>0)throw new Error("Positional arguments must come before keyword arguments");V.push(this.evaluate(oe,ee))}return[V,Y]}applyFilter($,ee,V){if(ee.type==="Identifier"){const Y=ee;if(Y.value==="tojson")return new we(kt($));if($ instanceof Te)switch(Y.value){case"list":return $;case"first":return $.value[0];case"last":return $.value[$.value.length-1];case"length":return new ye($.value.length);case"reverse":return new Te($.value.reverse());case"sort":return new Te($.value.sort((oe,xe)=>{if(oe.type!==xe.type)throw new Error(`Cannot compare different types: ${oe.type} and ${xe.type}`);switch(oe.type){case"IntegerValue":case"FloatValue":return oe.value-xe.value;case"StringValue":return oe.value.localeCompare(xe.value);default:throw new Error(`Cannot compare type: ${oe.type}`)}}));case"join":return new we($.value.map(oe=>oe.value).join(""));case"string":return new we(kt($));case"unique":{const oe=new Set,xe=[];for(const De of $.value)oe.has(De.value)||(oe.add(De.value),xe.push(De));return new Te(xe)}default:throw new Error(`Unknown ArrayValue filter: ${Y.value}`)}else if($ instanceof we)switch(Y.value){case"length":case"upper":case"lower":case"title":case"capitalize":{const oe=$.builtins.get(Y.value);if(oe instanceof qe)return oe.value([],V);if(oe instanceof ye)return oe;throw new Error(`Unknown StringValue filter: ${Y.value}`)}case"trim":return new we($.value.trim());case"indent":return new we($.value.split(`
`).map((oe,xe)=>xe===0||oe.length===0?oe:"    "+oe).join(`
`));case"join":case"string":return $;case"int":{const oe=parseInt($.value,10);return new ye(isNaN(oe)?0:oe)}case"float":{const oe=parseFloat($.value);return new de(isNaN(oe)?0:oe)}default:throw new Error(`Unknown StringValue filter: ${Y.value}`)}else if($ instanceof ye||$ instanceof de)switch(Y.value){case"abs":return $ instanceof ye?new ye(Math.abs($.value)):new de(Math.abs($.value));case"int":return new ye(Math.floor($.value));case"float":return new de($.value);default:throw new Error(`Unknown NumericValue filter: ${Y.value}`)}else if($ instanceof ke)switch(Y.value){case"items":return new Te(Array.from($.value.entries()).map(([oe,xe])=>new Te([new we(oe),xe])));case"length":return new ye($.value.size);default:throw new Error(`Unknown ObjectValue filter: ${Y.value}`)}else if($ instanceof ce)switch(Y.value){case"bool":return new ce($.value);case"int":return new ye($.value?1:0);case"float":return new de($.value?1:0);case"string":return new we($.value?"true":"false");default:throw new Error(`Unknown BooleanValue filter: ${Y.value}`)}throw new Error(`Cannot apply filter "${Y.value}" to type: ${$.type}`)}else if(ee.type==="CallExpression"){const Y=ee;if(Y.callee.type!=="Identifier")throw new Error(`Unknown filter: ${Y.callee.type}`);const oe=Y.callee.value;if(oe==="tojson"){const[,xe]=this.evaluateArguments(Y.args,V),De=xe.get("indent")??new st;if(!(De instanceof ye||De instanceof st))throw new Error("If set, indent must be a number");return new we(kt($,De.value))}else if(oe==="join"){let xe;if($ instanceof we)xe=Array.from($.value);else if($ instanceof Te)xe=$.value.map(pt=>pt.value);else throw new Error(`Cannot apply filter "${oe}" to type: ${$.type}`);const[De,nt]=this.evaluateArguments(Y.args,V),wt=De.at(0)??nt.get("separator")??new we("");if(!(wt instanceof we))throw new Error("separator must be a string");return new we(xe.join(wt.value))}else if(oe==="int"||oe==="float"){const[xe,De]=this.evaluateArguments(Y.args,V),nt=xe.at(0)??De.get("default")??(oe==="int"?new ye(0):new de(0));if($ instanceof we){const wt=oe==="int"?parseInt($.value,10):parseFloat($.value);return isNaN(wt)?nt:oe==="int"?new ye(wt):new de(wt)}else{if($ instanceof ye||$ instanceof de)return $;if($ instanceof ce)return oe==="int"?new ye($.value?1:0):new de($.value?1:0);throw new Error(`Cannot apply filter "${oe}" to type: ${$.type}`)}}else if(oe==="default"){const[xe,De]=this.evaluateArguments(Y.args,V),nt=xe[0]??new we(""),wt=xe[1]??De.get("boolean")??new ce(!1);if(!(wt instanceof ce))throw new Error("`default` filter flag must be a boolean");return $ instanceof Ze||wt.value&&!$.__bool__().value?nt:$}if($ instanceof Te){switch(oe){case"selectattr":case"rejectattr":{const xe=oe==="selectattr";if($.value.some(tt=>!(tt instanceof ke)))throw new Error(`\`${oe}\` can only be applied to array of objects`);if(Y.args.some(tt=>tt.type!=="StringLiteral"))throw new Error(`arguments of \`${oe}\` must be strings`);const[De,nt,wt]=Y.args.map(tt=>this.evaluate(tt,V));let pt;if(nt){const tt=V.tests.get(nt.value);if(!tt)throw new Error(`Unknown test: ${nt.value}`);pt=tt}else pt=(...tt)=>tt[0].__bool__().value;const xt=$.value.filter(tt=>{const It=tt.value.get(De.value),qt=It?pt(It,wt):!1;return xe?qt:!qt});return new Te(xt)}case"map":{const[,xe]=this.evaluateArguments(Y.args,V);if(xe.has("attribute")){const De=xe.get("attribute");if(!(De instanceof we))throw new Error("attribute must be a string");const nt=xe.get("default"),wt=$.value.map(pt=>{if(!(pt instanceof ke))throw new Error("items in map must be an object");return pt.value.get(De.value)??nt??new Ze});return new Te(wt)}else throw new Error("`map` expressions without `attribute` set are not currently supported.")}}throw new Error(`Unknown ArrayValue filter: ${oe}`)}else if($ instanceof we){switch(oe){case"indent":{const[xe,De]=this.evaluateArguments(Y.args,V),nt=xe.at(0)??De.get("width")??new ye(4);if(!(nt instanceof ye))throw new Error("width must be a number");const wt=xe.at(1)??De.get("first")??new ce(!1),pt=xe.at(2)??De.get("blank")??new ce(!1),xt=$.value.split(`
`),tt=" ".repeat(nt.value),It=xt.map((qt,Wr)=>!wt.value&&Wr===0||!pt.value&&qt.length===0?qt:tt+qt);return new we(It.join(`
`))}case"replace":{const xe=$.builtins.get("replace");if(!(xe instanceof qe))throw new Error("replace filter not available");const[De,nt]=this.evaluateArguments(Y.args,V);return xe.value([...De,new Le(nt)],V)}}throw new Error(`Unknown StringValue filter: ${oe}`)}else throw new Error(`Cannot apply filter "${oe}" to type: ${$.type}`)}throw new Error(`Unknown filter: ${ee.type}`)}evaluateFilterExpression($,ee){const V=this.evaluate($.operand,ee);return this.applyFilter(V,$.filter,ee)}evaluateTestExpression($,ee){const V=this.evaluate($.operand,ee),Y=ee.tests.get($.test.value);if(!Y)throw new Error(`Unknown test: ${$.test.value}`);const oe=Y(V);return new ce($.negate?!oe:oe)}evaluateSelectExpression($,ee){return this.evaluate($.test,ee).__bool__().value?this.evaluate($.lhs,ee):new Ze}evaluateUnaryExpression($,ee){const V=this.evaluate($.argument,ee);switch($.operator.value){case"not":return new ce(!V.value);default:throw new SyntaxError(`Unknown operator: ${$.operator.value}`)}}evaluateTernaryExpression($,ee){return this.evaluate($.condition,ee).__bool__().value?this.evaluate($.trueExpr,ee):this.evaluate($.falseExpr,ee)}evalProgram($,ee){return this.evaluateBlock($.body,ee)}evaluateBlock($,ee){let V="";for(const Y of $){const oe=this.evaluate(Y,ee);oe.type!=="NullValue"&&oe.type!=="UndefinedValue"&&(V+=oe.toString())}return new we(V)}evaluateIdentifier($,ee){return ee.lookupVariable($.value)}evaluateCallExpression($,ee){const[V,Y]=this.evaluateArguments($.args,ee);Y.size>0&&V.push(new Le(Y));const oe=this.evaluate($.callee,ee);if(oe.type!=="FunctionValue")throw new Error(`Cannot call something that is not a function: got ${oe.type}`);return oe.value(V,ee)}evaluateSliceExpression($,ee,V){if(!($ instanceof Te||$ instanceof we))throw new Error("Slice object must be an array or string");const Y=this.evaluate(ee.start,V),oe=this.evaluate(ee.stop,V),xe=this.evaluate(ee.step,V);if(!(Y instanceof ye||Y instanceof Ze))throw new Error("Slice start must be numeric or undefined");if(!(oe instanceof ye||oe instanceof Ze))throw new Error("Slice stop must be numeric or undefined");if(!(xe instanceof ye||xe instanceof Ze))throw new Error("Slice step must be numeric or undefined");return $ instanceof Te?new Te(Ae($.value,Y.value,oe.value,xe.value)):new we(Ae(Array.from($.value),Y.value,oe.value,xe.value).join(""))}evaluateMemberExpression($,ee){const V=this.evaluate($.object,ee);let Y;if($.computed){if($.property.type==="SliceExpression")return this.evaluateSliceExpression(V,$.property,ee);Y=this.evaluate($.property,ee)}else Y=new we($.property.value);let oe;if(V instanceof ke){if(!(Y instanceof we))throw new Error(`Cannot access property with non-string: got ${Y.type}`);oe=V.value.get(Y.value)??V.builtins.get(Y.value)}else if(V instanceof Te||V instanceof we)if(Y instanceof ye)oe=V.value.at(Y.value),V instanceof we&&(oe=new we(V.value.at(Y.value)));else if(Y instanceof we)oe=V.builtins.get(Y.value);else throw new Error(`Cannot access property with non-string/non-number: got ${Y.type}`);else{if(!(Y instanceof we))throw new Error(`Cannot access property with non-string: got ${Y.type}`);oe=V.builtins.get(Y.value)}return oe instanceof Ce?oe:new Ze}evaluateSet($,ee){const V=$.value?this.evaluate($.value,ee):this.evaluateBlock($.body,ee);if($.assignee.type==="Identifier"){const Y=$.assignee.value;ee.setVariable(Y,V)}else if($.assignee.type==="TupleLiteral"){const Y=$.assignee;if(!(V instanceof Te))throw new Error(`Cannot unpack non-iterable type in set: ${V.type}`);const oe=V.value;if(oe.length!==Y.value.length)throw new Error(`Too ${Y.value.length>oe.length?"few":"many"} items to unpack in set`);for(let xe=0;xe<Y.value.length;++xe){const De=Y.value[xe];if(De.type!=="Identifier")throw new Error(`Cannot unpack to non-identifier in set: ${De.type}`);ee.setVariable(De.value,oe[xe])}}else if($.assignee.type==="MemberExpression"){const Y=$.assignee,oe=this.evaluate(Y.object,ee);if(!(oe instanceof ke))throw new Error("Cannot assign to member of non-object");if(Y.property.type!=="Identifier")throw new Error("Cannot assign to member with non-identifier property");oe.value.set(Y.property.value,V)}else throw new Error(`Invalid LHS inside assignment expression: ${JSON.stringify($.assignee)}`);return new st}evaluateIf($,ee){const V=this.evaluate($.test,ee);return this.evaluateBlock(V.__bool__().value?$.body:$.alternate,ee)}evaluateFor($,ee){const V=new ze(ee);let Y,oe;if($.iterable.type==="SelectExpression"){const pt=$.iterable;oe=this.evaluate(pt.lhs,V),Y=pt.test}else oe=this.evaluate($.iterable,V);if(!(oe instanceof Te||oe instanceof ke))throw new Error(`Expected iterable or object type in for loop: got ${oe.type}`);oe instanceof ke&&(oe=oe.keys());const xe=[],De=[];for(let pt=0;pt<oe.value.length;++pt){const xt=new ze(V),tt=oe.value[pt];let It;if($.loopvar.type==="Identifier")It=qt=>qt.setVariable($.loopvar.value,tt);else if($.loopvar.type==="TupleLiteral"){const qt=$.loopvar;if(tt.type!=="ArrayValue")throw new Error(`Cannot unpack non-iterable type: ${tt.type}`);const Wr=tt;if(qt.value.length!==Wr.value.length)throw new Error(`Too ${qt.value.length>Wr.value.length?"few":"many"} items to unpack`);It=qr=>{for(let nr=0;nr<qt.value.length;++nr){if(qt.value[nr].type!=="Identifier")throw new Error(`Cannot unpack non-identifier type: ${qt.value[nr].type}`);qr.setVariable(qt.value[nr].value,Wr.value[nr])}}}else throw new Error(`Invalid loop variable(s): ${$.loopvar.type}`);Y&&(It(xt),!this.evaluate(Y,xt).__bool__().value)||(xe.push(tt),De.push(It))}let nt="",wt=!0;for(let pt=0;pt<xe.length;++pt){const xt=new Map([["index",new ye(pt+1)],["index0",new ye(pt)],["revindex",new ye(xe.length-pt)],["revindex0",new ye(xe.length-pt-1)],["first",new ce(pt===0)],["last",new ce(pt===xe.length-1)],["length",new ye(xe.length)],["previtem",pt>0?xe[pt-1]:new Ze],["nextitem",pt<xe.length-1?xe[pt+1]:new Ze]]);V.setVariable("loop",new ke(xt)),De[pt](V);try{const tt=this.evaluateBlock($.body,V);nt+=tt.value}catch(tt){if(tt instanceof Me)continue;if(tt instanceof Ee)break;throw tt}wt=!1}if(wt){const pt=this.evaluateBlock($.defaultBlock,V);nt+=pt.value}return new we(nt)}evaluateMacro($,ee){return ee.setVariable($.name.value,new qe((V,Y)=>{var De;const oe=new ze(Y);V=V.slice();let xe;((De=V.at(-1))==null?void 0:De.type)==="KeywordArgumentsValue"&&(xe=V.pop());for(let nt=0;nt<$.args.length;++nt){const wt=$.args[nt],pt=V[nt];if(wt.type==="Identifier"){const xt=wt;if(!pt)throw new Error(`Missing positional argument: ${xt.value}`);oe.setVariable(xt.value,pt)}else if(wt.type==="KeywordArgumentExpression"){const xt=wt,tt=pt??(xe==null?void 0:xe.value.get(xt.key.value))??this.evaluate(xt.value,oe);oe.setVariable(xt.key.value,tt)}else throw new Error(`Unknown argument type: ${wt.type}`)}return this.evaluateBlock($.body,oe)})),new st}evaluateCallStatement($,ee){const V=new qe((nt,wt)=>{const pt=new ze(wt);if($.callerArgs)for(let xt=0;xt<$.callerArgs.length;++xt){const tt=$.callerArgs[xt];if(tt.type!=="Identifier")throw new Error(`Caller parameter must be an identifier, got ${tt.type}`);pt.setVariable(tt.value,nt[xt]??new Ze)}return this.evaluateBlock($.body,pt)}),[Y,oe]=this.evaluateArguments($.call.args,ee);Y.push(new Le(oe));const xe=this.evaluate($.call.callee,ee);if(xe.type!=="FunctionValue")throw new Error(`Cannot call something that is not a function: got ${xe.type}`);const De=new ze(ee);return De.setVariable("caller",V),xe.value(Y,De)}evaluateFilterStatement($,ee){const V=this.evaluateBlock($.body,ee);return this.applyFilter(V,$.filter,ee)}evaluate($,ee){if(!$)return new Ze;switch($.type){case"Program":return this.evalProgram($,ee);case"Set":return this.evaluateSet($,ee);case"If":return this.evaluateIf($,ee);case"For":return this.evaluateFor($,ee);case"Macro":return this.evaluateMacro($,ee);case"CallStatement":return this.evaluateCallStatement($,ee);case"Break":throw new Ee;case"Continue":throw new Me;case"IntegerLiteral":return new ye($.value);case"FloatLiteral":return new de($.value);case"StringLiteral":return new we($.value);case"ArrayLiteral":return new Te($.value.map(V=>this.evaluate(V,ee)));case"TupleLiteral":return new We($.value.map(V=>this.evaluate(V,ee)));case"ObjectLiteral":{const V=new Map;for(const[Y,oe]of $.value){const xe=this.evaluate(Y,ee);if(!(xe instanceof we))throw new Error(`Object keys must be strings: got ${xe.type}`);V.set(xe.value,this.evaluate(oe,ee))}return new ke(V)}case"Identifier":return this.evaluateIdentifier($,ee);case"CallExpression":return this.evaluateCallExpression($,ee);case"MemberExpression":return this.evaluateMemberExpression($,ee);case"UnaryExpression":return this.evaluateUnaryExpression($,ee);case"BinaryExpression":return this.evaluateBinaryExpression($,ee);case"FilterExpression":return this.evaluateFilterExpression($,ee);case"FilterStatement":return this.evaluateFilterStatement($,ee);case"TestExpression":return this.evaluateTestExpression($,ee);case"SelectExpression":return this.evaluateSelectExpression($,ee);case"Ternary":return this.evaluateTernaryExpression($,ee);case"Comment":return new st;default:throw new SyntaxError(`Unknown node type: ${$.type}`)}}};function dt($){switch(typeof $){case"number":return Number.isInteger($)?new ye($):new de($);case"string":return new we($);case"boolean":return new ce($);case"undefined":return new Ze;case"object":return $===null?new st:Array.isArray($)?new Te($.map(dt)):new ke(new Map(Object.entries($).map(([ee,V])=>[ee,dt(V)])));case"function":return new qe((ee,V)=>{const Y=$(...ee.map(oe=>oe.value))??null;return dt(Y)});default:throw new Error(`Cannot convert to runtime value: ${$}`)}}function kt($,ee,V){const Y=V??0;switch($.type){case"NullValue":case"UndefinedValue":return"null";case"IntegerValue":case"FloatValue":case"StringValue":case"BooleanValue":return JSON.stringify($.value);case"ArrayValue":case"ObjectValue":{const oe=ee?" ".repeat(ee):"",xe=`
`+oe.repeat(Y),De=xe+oe;if($.type==="ArrayValue"){const nt=$.value.map(wt=>kt(wt,ee,Y+1));return ee?`[${De}${nt.join(`,${De}`)}${xe}]`:`[${nt.join(", ")}]`}else{const nt=Array.from($.value.entries()).map(([wt,pt])=>{const xt=`"${wt}": ${kt(pt,ee,Y+1)}`;return ee?`${De}${xt}`:xt});return ee?`{${nt.join(",")}${xe}}`:`{${nt.join(", ")}}`}}default:throw new Error(`Cannot convert to JSON: ${$.type}`)}}var ht=`
`,yr="{%- ",$r=" -%}";function Vr($){switch($.operator.type){case"MultiplicativeBinaryOperator":return 4;case"AdditiveBinaryOperator":return 3;case"ComparisonBinaryOperator":return 2;case"Identifier":return $.operator.value==="and"?1:$.operator.value==="in"||$.operator.value==="not in"?2:0}return 0}function Ur($,ee="	"){const V=typeof ee=="number"?" ".repeat(ee):ee;return Ar($.body,0,V).replace(/\n$/,"")}function sr(...$){return yr+$.join(" ")+$r}function Ar($,ee,V){return $.map(Y=>rn(Y,ee,V)).join(ht)}function rn($,ee,V){const Y=V.repeat(ee);switch($.type){case"Program":return Ar($.body,ee,V);case"If":return sn($,ee,V);case"For":return nn($,ee,V);case"Set":return ds($,ee,V);case"Macro":return ft($,ee,V);case"Break":return Y+sr("break");case"Continue":return Y+sr("continue");case"CallStatement":return Os($,ee,V);case"FilterStatement":return Ds($,ee,V);case"Comment":return Y+"{# "+$.value+" #}";default:return Y+"{{- "+St($)+" -}}"}}function sn($,ee,V){const Y=V.repeat(ee),oe=[];let xe=$;for(;xe&&(oe.push({test:xe.test,body:xe.body}),xe.alternate.length===1&&xe.alternate[0].type==="If");)xe=xe.alternate[0];let De=Y+sr("if",St(oe[0].test))+ht+Ar(oe[0].body,ee+1,V);for(let nt=1;nt<oe.length;++nt)De+=ht+Y+sr("elif",St(oe[nt].test))+ht+Ar(oe[nt].body,ee+1,V);return xe&&xe.alternate.length>0&&(De+=ht+Y+sr("else")+ht+Ar(xe.alternate,ee+1,V)),De+=ht+Y+sr("endif"),De}function nn($,ee,V){const Y=V.repeat(ee);let oe="";if($.iterable.type==="SelectExpression"){const De=$.iterable;oe=`${St(De.lhs)} if ${St(De.test)}`}else oe=St($.iterable);let xe=Y+sr("for",St($.loopvar),"in",oe)+ht+Ar($.body,ee+1,V);return $.defaultBlock.length>0&&(xe+=ht+Y+sr("else")+ht+Ar($.defaultBlock,ee+1,V)),xe+=ht+Y+sr("endfor"),xe}function ds($,ee,V){const Y=V.repeat(ee),oe=St($.assignee),xe=$.value?St($.value):"",De=Y+sr("set",`${oe}${$.value?" = "+xe:""}`);return $.body.length===0?De:De+ht+Ar($.body,ee+1,V)+ht+Y+sr("endset")}function ft($,ee,V){const Y=V.repeat(ee),oe=$.args.map(St).join(", ");return Y+sr("macro",`${$.name.value}(${oe})`)+ht+Ar($.body,ee+1,V)+ht+Y+sr("endmacro")}function Os($,ee,V){const Y=V.repeat(ee),oe=$.callerArgs&&$.callerArgs.length>0?`(${$.callerArgs.map(St).join(", ")})`:"",xe=St($.call);let De=Y+sr(`call${oe}`,xe)+ht;return De+=Ar($.body,ee+1,V)+ht,De+=Y+sr("endcall"),De}function Ds($,ee,V){const Y=V.repeat(ee),oe=$.filter.type==="Identifier"?$.filter.value:St($.filter);let xe=Y+sr("filter",oe)+ht;return xe+=Ar($.body,ee+1,V)+ht,xe+=Y+sr("endfilter"),xe}function St($,ee=-1){switch($.type){case"SpreadExpression":return`*${St($.argument)}`;case"Identifier":return $.value;case"IntegerLiteral":return`${$.value}`;case"FloatLiteral":return`${$.value}`;case"StringLiteral":return JSON.stringify($.value);case"BinaryExpression":{const V=$,Y=Vr(V),oe=St(V.left,Y),xe=St(V.right,Y+1),De=`${oe} ${V.operator.value} ${xe}`;return Y<ee?`(${De})`:De}case"UnaryExpression":{const V=$;return V.operator.value+(V.operator.value==="not"?" ":"")+St(V.argument,1/0)}case"CallExpression":{const V=$,Y=V.args.map(St).join(", ");return`${St(V.callee)}(${Y})`}case"MemberExpression":{const V=$;let Y=St(V.object);["Identifier","MemberExpression","CallExpression","StringLiteral","IntegerLiteral","FloatLiteral","ArrayLiteral","TupleLiteral","ObjectLiteral"].includes(V.object.type)||(Y=`(${Y})`);let oe=St(V.property);return!V.computed&&V.property.type!=="Identifier"&&(oe=`(${oe})`),V.computed?`${Y}[${oe}]`:`${Y}.${oe}`}case"FilterExpression":{const V=$,Y=St(V.operand,1/0);return V.filter.type==="CallExpression"?`${Y} | ${St(V.filter)}`:`${Y} | ${V.filter.value}`}case"SelectExpression":{const V=$;return`${St(V.lhs)} if ${St(V.test)}`}case"TestExpression":{const V=$;return`${St(V.operand)} is${V.negate?" not":""} ${V.test.value}`}case"ArrayLiteral":case"TupleLiteral":{const V=$.value.map(St),Y=$.type==="ArrayLiteral"?"[]":"()";return`${Y[0]}${V.join(", ")}${Y[1]}`}case"ObjectLiteral":return`{${Array.from($.value.entries()).map(([Y,oe])=>`${St(Y)}: ${St(oe)}`).join(", ")}}`;case"SliceExpression":{const V=$,Y=V.start?St(V.start):"",oe=V.stop?St(V.stop):"",xe=V.step?`:${St(V.step)}`:"";return`${Y}:${oe}${xe}`}case"KeywordArgumentExpression":{const V=$;return`${V.key.value}=${St(V.value)}`}case"Ternary":{const V=$,Y=`${St(V.trueExpr)} if ${St(V.condition,0)} else ${St(V.falseExpr)}`;return ee>-1?`(${Y})`:Y}default:throw new Error(`Unknown expression type: ${$.type}`)}}var Kt=class{constructor($){re(this,"parsed");const ee=p($,{lstrip_blocks:!0,trim_blocks:!0});this.parsed=ie(ee)}render($){const ee=new ze;if(He(ee),$)for(const[oe,xe]of Object.entries($))ee.set(oe,xe);return new gt(ee).run(this.parsed).value}format($){return Ur(this.parsed,($==null?void 0:$.indent)||"	")}}},"./src/backends/onnx.js":(e,r,t)=>{var s;t.r(r),t.d(r,{Tensor:()=>i.Tensor,createInferenceSession:()=>A,deviceToExecutionProviders:()=>f,isONNXProxy:()=>T,isONNXTensor:()=>x,runInferenceSession:()=>C});var o=t("./src/env.js"),n=t("?2ce3"),a=t("onnxruntime-web"),i=t("onnxruntime-common");const l=Object.freeze({auto:null,gpu:null,cpu:"cpu",wasm:"wasm",webgpu:"webgpu",cuda:"cuda",dml:"dml",webnn:{name:"webnn",deviceType:"cpu"},"webnn-npu":{name:"webnn",deviceType:"npu"},"webnn-gpu":{name:"webnn",deviceType:"gpu"},"webnn-cpu":{name:"webnn",deviceType:"cpu"}}),c=[];let p,d;const u=Symbol.for("onnxruntime");if(u in globalThis)d=globalThis[u];else if(o.apis.IS_NODE_ENV){switch(d=n??(s||(s=t.t(n,2))),process.platform){case"win32":c.push("dml");break;case"linux":process.arch==="x64"&&c.push("cuda");break}c.push("cpu"),p=["cpu"]}else d=a,o.apis.IS_WEBNN_AVAILABLE&&c.push("webnn-npu","webnn-gpu","webnn-cpu","webnn"),o.apis.IS_WEBGPU_AVAILABLE&&c.push("webgpu"),c.push("wasm"),p=["wasm"];const _=d.InferenceSession;function f(v=null){if(!v)return p;switch(v){case"auto":return c;case"gpu":return c.filter(P=>["webgpu","cuda","dml","webnn-gpu"].includes(P))}if(c.includes(v))return[l[v]??v];throw new Error(`Unsupported device: "${v}". Should be one of: ${c.join(", ")}.`)}let b=null;async function A(v,P,F){b&&await b;const D=_.create(v,P);b??(b=D);const K=await D;return K.config=F,K}let g=Promise.resolve();const y=o.apis.IS_BROWSER_ENV||o.apis.IS_WEBWORKER_ENV;async function C(v,P){const F=()=>v.run(P);return await(y?g=g.then(F):F())}function x(v){return v instanceof d.Tensor}const M=d==null?void 0:d.env;M!=null&&M.wasm&&(!(typeof ServiceWorkerGlobalScope<"u"&&self instanceof ServiceWorkerGlobalScope)&&!M.wasm.wasmPaths&&(M.wasm.wasmPaths=`https://cdn.jsdelivr.net/npm/@huggingface/transformers@${o.env.version}/dist/`),M.wasm.proxy=!1),M!=null&&M.webgpu&&(M.webgpu.powerPreference="high-performance");function T(){var v;return(v=M==null?void 0:M.wasm)==null?void 0:v.proxy}o.env.backends.onnx=M},"./src/base/feature_extraction_utils.js":(e,r,t)=>{t.r(r),t.d(r,{FeatureExtractor:()=>a,validate_audio_inputs:()=>i});var s=t("./src/utils/constants.js"),o=t("./src/utils/generic.js"),n=t("./src/utils/hub.js");class a extends o.Callable{constructor(c){super(),this.config=c}static async from_pretrained(c,p={}){const d=await(0,n.getModelJSON)(c,s.FEATURE_EXTRACTOR_NAME,!0,p);return new this(d)}}function i(l,c){var p;if(!(l instanceof Float32Array||l instanceof Float64Array))throw new Error(`${c} expects input to be a Float32Array or a Float64Array, but got ${((p=l==null?void 0:l.constructor)==null?void 0:p.name)??typeof l} instead. If using the feature extractor directly, remember to use \`read_audio(url, sampling_rate)\` to obtain the raw audio data of the file/url.`)}},"./src/base/image_processors_utils.js":(e,r,t)=>{t.r(r),t.d(r,{ImageProcessor:()=>x,center_to_corners_format:()=>d,post_process_instance_segmentation:()=>C,post_process_object_detection:()=>u,post_process_panoptic_segmentation:()=>y,post_process_semantic_segmentation:()=>_});var s=t("./src/utils/generic.js"),o=t("./src/utils/tensor.js"),n=t("./src/utils/maths.js");t("./src/utils/image.js");var a=t("./src/utils/core.js"),i=t("./src/utils/hub.js"),l=t("./src/utils/constants.js");function c(M,T,v=0,P=null){const F=M/T;let D=(0,n.bankers_round)(F)*T;return P!==null&&D>P&&(D=Math.floor(F)*T),D<v&&(D=Math.ceil(F)*T),D}function p([M,T],v){return[Math.max(Math.floor(M/v),1)*v,Math.max(Math.floor(T/v),1)*v]}function d([M,T,v,P]){return[M-v/2,T-P/2,M+v/2,T+P/2]}function u(M,T=.5,v=null,P=!1){const F=M.logits,D=M.pred_boxes,[K,U,j]=F.dims;if(v!==null&&v.length!==K)throw Error("Make sure that you pass in as many target sizes as the batch dimension of the logits");let ne=[];for(let q=0;q<K;++q){let te=v!==null?v[q]:null,Z={boxes:[],classes:[],scores:[]},ae=F[q],he=D[q];for(let Q=0;Q<U;++Q){let B=ae[Q],O=[],W;if(P){W=B.sigmoid().data;for(let N=0;N<W.length;++N)W[N]>T&&O.push(N)}else{let N=(0,n.max)(B.data)[1];if(N===j-1||(W=(0,n.softmax)(B.data),W[N]<T))continue;O.push(N)}for(const N of O){let J=he[Q].data;J=d(J),te!==null&&(J=J.map((ie,me)=>ie*te[(me+1)%2])),Z.boxes.push(J),Z.classes.push(N),Z.scores.push(W[N])}}ne.push(Z)}return ne}function _(M,T=null){const v=M.logits,P=v.dims[0];if(T!==null&&T.length!==P)throw Error("Make sure that you pass in as many target sizes as the batch dimension of the logits");const F=[];for(let D=0;D<P;++D){const K=T!==null?T[D]:null;let U=v[D];K!==null&&(U=(0,o.interpolate)(U,K,"bilinear",!1));const[j,ne]=K??U.dims.slice(-2),q=new o.Tensor("int32",new Int32Array(j*ne),[j,ne]),te=U[0].data,Z=q.data;for(let Q=1;Q<U.dims[0];++Q){const B=U[Q].data;for(let O=0;O<B.length;++O)B[O]>te[O]&&(te[O]=B[O],Z[O]=Q)}const ae=new Array(U.dims[0]);for(let Q=0;Q<Z.length;++Q){const B=Z[Q];ae[B]=B}const he=ae.filter(Q=>Q!==void 0);F.push({segmentation:q,labels:he})}return F}function f(M,T,v,P){const F=[],D=[],K=[];for(let U=0;U<M.dims[0];++U){const j=M[U],ne=T[U],q=(0,n.max)(j.data)[1];if(q===P)continue;const Z=(0,n.softmax)(j.data)[q];Z>v&&(F.push(ne),D.push(Z),K.push(q))}return[F,D,K]}function b(M,T,v,P=.5,F=.8){const D=[];let K=0,U=0;const j=T[v].data;for(let q=0;q<M.length;++q)M[q]===v&&(D.push(q),++K),j[q]>=P&&++U;let ne=K>0&&U>0;return ne&&(ne=K/U>F),[ne,D]}function A(M,T,v,P,F,D=null,K=null){const[U,j]=K??M[0].dims,ne=new o.Tensor("int32",new Int32Array(U*j),[U,j]),q=[];if(K!==null)for(let Q=0;Q<M.length;++Q)M[Q]=(0,o.interpolate)(M[Q],K,"bilinear",!1);const te=new Int32Array(M[0].data.length),Z=new Float32Array(M[0].data.length);for(let Q=0;Q<M.length;++Q){let B=T[Q];const O=M[Q].data;for(let W=0;W<O.length;++W)O[W]*=B,O[W]>Z[W]&&(te[W]=Q,Z[W]=O[W])}let ae=0;const he=ne.data;for(let Q=0;Q<v.length;++Q){const B=v[Q],[O,W]=b(te,M,Q,P,F);if(O){++ae;for(const N of W)he[N]=ae;q.push({id:ae,label_id:B,score:T[Q]})}}return[ne,q]}function g(M,T,v=28,P=3136,F=784*1280){if(M<v||T<v)throw new Error(`height:${M} or width:${T} must be larger than factor:${v}`);if(Math.max(M,T)/Math.min(M,T)>200)throw new Error(`absolute aspect ratio must be smaller than 200, got ${Math.max(M,T)/Math.min(M,T)}`);let D=Math.round(M/v)*v,K=Math.round(T/v)*v;if(D*K>F){const U=Math.sqrt(M*T/F);D=Math.floor(M/U/v)*v,K=Math.floor(T/U/v)*v}else if(D*K<P){const U=Math.sqrt(P/(M*T));D=Math.ceil(M*U/v)*v,K=Math.ceil(T*U/v)*v}return[D,K]}function y(M,T=.5,v=.5,P=.8,F=null,D=null){F===null&&(console.warn("`label_ids_to_fuse` unset. No instance will be fused."),F=new Set);const K=M.class_queries_logits??M.logits,j=(M.masks_queries_logits??M.pred_masks).sigmoid();let[ne,q,te]=K.dims;if(te-=1,D!==null&&D.length!==ne)throw Error("Make sure that you pass in as many target sizes as the batch dimension of the logits");let Z=[];for(let ae=0;ae<ne;++ae){let he=D!==null?D[ae]:null,Q=K[ae],B=j[ae],[O,W,N]=f(Q,B,T,te);if(N.length===0){let[me,Ae]=he??B.dims.slice(-2),Ve=new o.Tensor("int32",new Int32Array(me*Ae).fill(-1),[me,Ae]);Z.push({segmentation:Ve,segments_info:[]});continue}let[J,ie]=A(O,W,N,v,P,F,he);Z.push({segmentation:J,segments_info:ie})}return Z}function C(M,T=.5,v=null){throw new Error("`post_process_instance_segmentation` is not yet implemented.")}class x extends s.Callable{constructor(T){super(),this.image_mean=T.image_mean??T.mean,this.image_std=T.image_std??T.std,this.resample=T.resample??2,this.do_rescale=T.do_rescale??!0,this.rescale_factor=T.rescale_factor??1/255,this.do_normalize=T.do_normalize,this.do_thumbnail=T.do_thumbnail,this.size=T.size??T.image_size,this.do_resize=T.do_resize??this.size!==void 0,this.size_divisibility=T.size_divisibility??T.size_divisor,this.do_center_crop=T.do_center_crop,this.crop_size=T.crop_size,this.do_convert_rgb=T.do_convert_rgb??!0,this.do_crop_margin=T.do_crop_margin,this.pad_size=T.pad_size,this.do_pad=T.do_pad,this.min_pixels=T.min_pixels,this.max_pixels=T.max_pixels,this.do_pad&&!this.pad_size&&this.size&&this.size.width!==void 0&&this.size.height!==void 0&&(this.pad_size=this.size),this.do_flip_channel_order=T.do_flip_channel_order??!1,this.config=T}async thumbnail(T,v,P=2){const F=T.height,D=T.width,K=v.height,U=v.width;let j=Math.min(F,K),ne=Math.min(D,U);return j===F&&ne===D?T:(F>D?ne=Math.floor(D*j/F):D>F&&(j=Math.floor(F*ne/D)),await T.resize(ne,j,{resample:P}))}async crop_margin(T,v=200){const P=T.clone().grayscale(),F=(0,n.min)(P.data)[0],K=(0,n.max)(P.data)[0]-F;if(K===0)return T;const U=v/255;let j=P.width,ne=P.height,q=0,te=0;const Z=P.data;for(let ae=0;ae<P.height;++ae){const he=ae*P.width;for(let Q=0;Q<P.width;++Q)(Z[he+Q]-F)/K<U&&(j=Math.min(j,Q),ne=Math.min(ne,ae),q=Math.max(q,Q),te=Math.max(te,ae))}return T=await T.crop([j,ne,q,te]),T}pad_image(T,v,P,{mode:F="constant",center:D=!1,constant_values:K=0}={}){const[U,j,ne]=v;let q,te;if(typeof P=="number"?(q=P,te=P):P==="square"?q=te=Math.max(U,j):(q=P.width,te=P.height),q!==j||te!==U){const Z=new Float32Array(q*te*ne);if(Array.isArray(K))for(let Q=0;Q<Z.length;++Q)Z[Q]=K[Q%ne];else K!==0&&Z.fill(K);const[ae,he]=D?[Math.floor((q-j)/2),Math.floor((te-U)/2)]:[0,0];for(let Q=0;Q<U;++Q){const B=(Q+he)*q,O=Q*j;for(let W=0;W<j;++W){const N=(B+W+ae)*ne,J=(O+W)*ne;for(let ie=0;ie<ne;++ie)Z[N+ie]=T[J+ie]}}if(F==="symmetric"){if(D)throw new Error("`center` padding is not supported when `mode` is set to `symmetric`.");const Q=U-1,B=j-1;for(let O=0;O<te;++O){const W=O*q,N=(0,a.calculateReflectOffset)(O,Q)*j;for(let J=0;J<q;++J){if(O<U&&J<j)continue;const ie=(W+J)*ne,me=(N+(0,a.calculateReflectOffset)(J,B))*ne;for(let Ae=0;Ae<ne;++Ae)Z[ie+Ae]=T[me+Ae]}}}T=Z,v=[te,q,ne]}return[T,v]}rescale(T){for(let v=0;v<T.length;++v)T[v]=this.rescale_factor*T[v]}get_resize_output_image_size(T,v){const[P,F]=T.size;let D,K;if(this.do_thumbnail){const{height:U,width:j}=v;D=Math.min(U,j)}else Number.isInteger(v)?(D=v,K=this.config.max_size??D):v!==void 0&&(D=v.shortest_edge,K=v.longest_edge);if(D!==void 0||K!==void 0){const U=D===void 0?1:Math.max(D/P,D/F),j=P*U,ne=F*U,q=K===void 0?1:Math.min(K/j,K/ne);let te=Math.floor(Number((j*q).toFixed(2))),Z=Math.floor(Number((ne*q).toFixed(2)));return this.size_divisibility!==void 0&&([te,Z]=p([te,Z],this.size_divisibility)),[te,Z]}else if(v!==void 0&&v.width!==void 0&&v.height!==void 0){let U=v.width,j=v.height;if(this.config.keep_aspect_ratio&&this.config.ensure_multiple_of){let ne=j/F,q=U/P;Math.abs(1-q)<Math.abs(1-ne)?ne=q:q=ne,j=c(ne*F,this.config.ensure_multiple_of),U=c(q*P,this.config.ensure_multiple_of)}return[U,j]}else{if(this.size_divisibility!==void 0)return p([P,F],this.size_divisibility);if(this.min_pixels!==void 0&&this.max_pixels!==void 0){const U=this.config.patch_size*this.config.merge_size;return g(F,P,U,this.min_pixels,this.max_pixels)}else throw new Error(`Could not resize image due to unsupported \`this.size\` option in config: ${JSON.stringify(v)}`)}}async resize(T){const[v,P]=this.get_resize_output_image_size(T,this.size);return await T.resize(v,P,{resample:this.resample})}async preprocess(T,{do_normalize:v=null,do_pad:P=null,do_convert_rgb:F=null,do_convert_grayscale:D=null,do_flip_channel_order:K=null}={}){this.do_crop_margin&&(T=await this.crop_margin(T));const[U,j]=T.size;if(F??this.do_convert_rgb?T=T.rgb():D&&(T=T.grayscale()),this.do_resize&&(T=await this.resize(T)),this.do_thumbnail&&(T=await this.thumbnail(T,this.size,this.resample)),this.do_center_crop){let ae,he;Number.isInteger(this.crop_size)?(ae=this.crop_size,he=this.crop_size):(ae=this.crop_size.width,he=this.crop_size.height),T=await T.center_crop(ae,he)}const ne=[T.height,T.width];let q=Float32Array.from(T.data),te=[T.height,T.width,T.channels];if(this.do_rescale&&this.rescale(q),v??this.do_normalize){let ae=this.image_mean;Array.isArray(this.image_mean)||(ae=new Array(T.channels).fill(ae));let he=this.image_std;if(Array.isArray(this.image_std)||(he=new Array(T.channels).fill(ae)),ae.length!==T.channels||he.length!==T.channels)throw new Error(`When set to arrays, the length of \`image_mean\` (${ae.length}) and \`image_std\` (${he.length}) must match the number of channels in the image (${T.channels}).`);for(let Q=0;Q<q.length;Q+=T.channels)for(let B=0;B<T.channels;++B)q[Q+B]=(q[Q+B]-ae[B])/he[B]}if(P??this.do_pad){if(this.pad_size)[q,te]=this.pad_image(q,[T.height,T.width,T.channels],this.pad_size);else if(this.size_divisibility){const[ae,he]=p([te[1],te[0]],this.size_divisibility);[q,te]=this.pad_image(q,te,{width:ae,height:he})}}if(K??this.do_flip_channel_order){if(te[2]!==3)throw new Error("Flipping channel order is only supported for RGB images.");for(let ae=0;ae<q.length;ae+=3){const he=q[ae];q[ae]=q[ae+2],q[ae+2]=he}}const Z=new o.Tensor("float32",q,te).permute(2,0,1);return{original_size:[j,U],reshaped_input_size:ne,pixel_values:Z}}async _call(T,...v){Array.isArray(T)||(T=[T]);const P=await Promise.all(T.map(D=>this.preprocess(D)));return{pixel_values:(0,o.stack)(P.map(D=>D.pixel_values),0),original_sizes:P.map(D=>D.original_size),reshaped_input_sizes:P.map(D=>D.reshaped_input_size)}}static async from_pretrained(T,v={}){const P=await(0,i.getModelJSON)(T,l.IMAGE_PROCESSOR_NAME,!0,v);return new this(P)}}},"./src/base/processing_utils.js":(e,r,t)=>{t.r(r),t.d(r,{Processor:()=>a});var s=t("./src/utils/constants.js"),o=t("./src/utils/generic.js"),n=t("./src/utils/hub.js");class a extends o.Callable{constructor(l,c,p){super(),this.config=l,this.components=c,this.chat_template=p}get image_processor(){return this.components.image_processor}get tokenizer(){return this.components.tokenizer}get feature_extractor(){return this.components.feature_extractor}apply_chat_template(l,c={}){if(!this.tokenizer)throw new Error("Unable to apply chat template without a tokenizer.");return this.tokenizer.apply_chat_template(l,{tokenize:!1,chat_template:this.chat_template??void 0,...c})}batch_decode(...l){if(!this.tokenizer)throw new Error("Unable to decode without a tokenizer.");return this.tokenizer.batch_decode(...l)}decode(...l){if(!this.tokenizer)throw new Error("Unable to decode without a tokenizer.");return this.tokenizer.decode(...l)}async _call(l,...c){for(const p of[this.image_processor,this.feature_extractor,this.tokenizer])if(p)return p(l,...c);throw new Error("No image processor, feature extractor, or tokenizer found.")}static async from_pretrained(l,c={}){const[p,d,u]=await Promise.all([this.uses_processor_config?(0,n.getModelJSON)(l,s.PROCESSOR_NAME,!0,c):{},Promise.all(this.classes.filter(_=>_ in this).map(async _=>{const f=await this[_].from_pretrained(l,c);return[_.replace(/_class$/,""),f]})).then(Object.fromEntries),this.uses_chat_template_file?(0,n.getModelText)(l,s.CHAT_TEMPLATE_NAME,!0,c):null]);return new this(p,d,u)}}re(a,"classes",["image_processor_class","tokenizer_class","feature_extractor_class"]),re(a,"uses_processor_config",!1),re(a,"uses_chat_template_file",!1)},"./src/configs.js":(e,r,t)=>{t.r(r),t.d(r,{AutoConfig:()=>p,PretrainedConfig:()=>c,getCacheShapes:()=>i});var s=t("./src/utils/core.js"),o=t("./src/utils/hub.js");async function n(d,u){return await(0,o.getModelJSON)(d,"config.json",!0,u)}function a(d){const u={};let _={};switch(d.model_type){case"llava":case"paligemma":case"gemma3":case"florence2":case"llava_onevision":case"idefics3":case"ultravox":case"voxtral":case"smolvlm":case"gemma3n":_=a(d.text_config);break;case"moondream1":_=a(d.phi_config);break;case"musicgen":_=a(d.decoder);break;case"multi_modality":_=a(d.language_config);break;case"gpt2":case"gptj":case"jais":case"codegen":case"gpt_bigcode":u.num_heads="n_head",u.num_layers="n_layer",u.hidden_size="n_embd";break;case"gpt_neox":case"stablelm":case"opt":case"falcon":case"modernbert-decoder":u.num_heads="num_attention_heads",u.num_layers="num_hidden_layers",u.hidden_size="hidden_size";break;case"llama":case"llama4_text":case"nanochat":case"arcee":case"lfm2":case"smollm3":case"olmo":case"olmo2":case"mobilellm":case"granite":case"granitemoehybrid":case"cohere":case"mistral":case"starcoder2":case"qwen2":case"qwen2_vl":case"phi":case"phi3":case"phi3_v":case"llava_qwen2":u.num_heads="num_key_value_heads",u.num_layers="num_hidden_layers",u.hidden_size="hidden_size",u.num_attention_heads="num_attention_heads",u.dim_kv="head_dim";break;case"qwen3":case"gemma":case"gemma2":case"vaultgemma":case"gemma3_text":case"gemma3n_text":case"glm":case"helium":case"ernie4_5":u.num_heads="num_key_value_heads",u.num_layers="num_hidden_layers",u.dim_kv="head_dim";break;case"openelm":u.num_heads="num_kv_heads",u.num_layers="num_transformer_layers",u.dim_kv="head_dim";break;case"gpt_neo":case"donut-swin":u.num_heads="num_heads",u.num_layers="num_layers",u.hidden_size="hidden_size";break;case"bloom":u.num_heads="n_head",u.num_layers="n_layer",u.hidden_size="hidden_size";break;case"mpt":u.num_heads="n_heads",u.num_layers="n_layers",u.hidden_size="d_model";break;case"exaone":u.num_heads="num_key_value_heads",u.num_layers="num_layers",u.dim_kv="head_dim",u.num_attention_heads="num_attention_heads";break;case"t5":case"mt5":case"longt5":u.num_decoder_layers="num_decoder_layers",u.num_decoder_heads="num_heads",u.decoder_dim_kv="d_kv",u.num_encoder_layers="num_layers",u.num_encoder_heads="num_heads",u.encoder_dim_kv="d_kv";break;case"bart":case"mbart":case"marian":case"whisper":case"lite-whisper":case"m2m_100":case"blenderbot":case"blenderbot-small":case"florence2_language":u.num_decoder_layers="decoder_layers",u.num_decoder_heads="decoder_attention_heads",u.decoder_hidden_size="d_model",u.num_encoder_layers="encoder_layers",u.num_encoder_heads="encoder_attention_heads",u.encoder_hidden_size="d_model";break;case"speecht5":u.num_decoder_layers="decoder_layers",u.num_decoder_heads="decoder_attention_heads",u.decoder_hidden_size="hidden_size",u.num_encoder_layers="encoder_layers",u.num_encoder_heads="encoder_attention_heads",u.encoder_hidden_size="hidden_size";break;case"trocr":u.num_encoder_layers=u.num_decoder_layers="decoder_layers",u.num_encoder_heads=u.num_decoder_heads="decoder_attention_heads",u.encoder_hidden_size=u.decoder_hidden_size="d_model";break;case"musicgen_decoder":u.num_encoder_layers=u.num_decoder_layers="num_hidden_layers",u.num_encoder_heads=u.num_decoder_heads="num_attention_heads",u.encoder_hidden_size=u.decoder_hidden_size="hidden_size";break;case"moonshine":u.num_decoder_layers="decoder_num_hidden_layers",u.num_decoder_heads="decoder_num_key_value_heads",u.num_encoder_layers="encoder_num_hidden_layers",u.num_encoder_heads="encoder_num_key_value_heads",u.encoder_hidden_size=u.decoder_hidden_size="hidden_size";break;case"vision-encoder-decoder":const b=a(d.decoder),A="num_decoder_layers"in b,g=(0,s.pick)(d,["model_type","is_encoder_decoder"]);return A?(g.num_decoder_layers=b.num_decoder_layers,g.num_decoder_heads=b.num_decoder_heads,g.decoder_hidden_size=b.decoder_hidden_size,g.num_encoder_layers=b.num_encoder_layers,g.num_encoder_heads=b.num_encoder_heads,g.encoder_hidden_size=b.encoder_hidden_size):(g.num_layers=b.num_layers,g.num_heads=b.num_heads,g.hidden_size=b.hidden_size),g}const f={..._,...(0,s.pick)(d,["model_type","multi_query","is_encoder_decoder"])};for(const b in u)f[b]=d[u[b]];return f}function i(d,u){if(d.model_type==="lfm2"){const _=(u==null?void 0:u.prefix)??"past_key_values",f=_==="present"?"present":"past",b={},{layer_types:A,num_attention_heads:g,num_key_value_heads:y,hidden_size:C,conv_L_cache:x}=d,M=C/g,T=(u==null?void 0:u.batch_size)??1;for(let v=0;v<A.length;++v)if(A[v]==="full_attention")for(const P of["key","value"])b[`${_}.${v}.${P}`]=[T,y,0,M];else if(A[v]==="conv")b[`${f}_conv.${v}`]=[T,C,x];else throw new Error(`Unsupported layer type: ${A[v]}`);return b}return l(d,u)}function l(d,{prefix:u="past_key_values",batch_size:_=1}={}){const f={},b=d.normalized_config;if(b.is_encoder_decoder&&"num_encoder_heads"in b&&"num_decoder_heads"in b){const A=b.encoder_dim_kv??b.encoder_hidden_size/b.num_encoder_heads,g=b.decoder_dim_kv??b.decoder_hidden_size/b.num_decoder_heads,y=[_,b.num_encoder_heads,0,A],C=[_,b.num_decoder_heads,0,g];for(let x=0;x<b.num_decoder_layers;++x)f[`${u}.${x}.encoder.key`]=y,f[`${u}.${x}.encoder.value`]=y,f[`${u}.${x}.decoder.key`]=C,f[`${u}.${x}.decoder.value`]=C}else{const A=b.num_heads,g=b.num_layers,y=b.dim_kv??b.hidden_size/(b.num_attention_heads??A);if(b.model_type==="falcon"){const C=[_*A,0,y];for(let x=0;x<g;++x)f[`${u}.${x}.key`]=C,f[`${u}.${x}.value`]=C}else if(b.multi_query){const C=[_*A,0,2*y];for(let x=0;x<g;++x)f[`${u}.${x}.key_value`]=C}else if(b.model_type==="bloom"){const C=[_*A,y,0],x=[_*A,0,y];for(let M=0;M<g;++M)f[`${u}.${M}.key`]=C,f[`${u}.${M}.value`]=x}else if(b.model_type==="openelm")for(let C=0;C<g;++C){const x=[_,A[C],0,y];f[`${u}.${C}.key`]=x,f[`${u}.${C}.value`]=x}else{const C=[_,A,0,y];for(let x=0;x<g;++x)f[`${u}.${x}.key`]=C,f[`${u}.${x}.value`]=C}}return f}class c{constructor(u){re(this,"model_type",null);re(this,"is_encoder_decoder",!1);re(this,"max_position_embeddings");re(this,"transformers.js_config");Object.assign(this,u),this.normalized_config=a(this)}static async from_pretrained(u,{progress_callback:_=null,config:f=null,cache_dir:b=null,local_files_only:A=!1,revision:g="main"}={}){f&&!(f instanceof c)&&(f=new c(f));const y=f??await n(u,{progress_callback:_,config:f,cache_dir:b,local_files_only:A,revision:g});return new this(y)}}class p{static async from_pretrained(...u){return c.from_pretrained(...u)}}},"./src/env.js":(e,r,t)=>{var F,D;t.r(r),t.d(r,{apis:()=>g,env:()=>v});var s=t("?db59"),o=t("?383f"),n=t("?fa4b");const a="3.7.6",i=typeof window<"u"&&typeof window.document<"u",l=typeof self<"u"&&["DedicatedWorkerGlobalScope","ServiceWorkerGlobalScope","SharedWorkerGlobalScope"].includes((F=self.constructor)==null?void 0:F.name),c=typeof self<"u"&&"caches"in self,p=typeof navigator<"u"&&"gpu"in navigator,d=typeof navigator<"u"&&"ml"in navigator,u=typeof process<"u",_=u&&((D=process==null?void 0:process.release)==null?void 0:D.name)==="node",f=!P(s),b=!P(o),A=typeof globalThis.Deno<"u",g=Object.freeze({IS_BROWSER_ENV:i,IS_WEBWORKER_ENV:l,IS_WEB_CACHE_AVAILABLE:c,IS_WEBGPU_AVAILABLE:p,IS_WEBNN_AVAILABLE:d,IS_PROCESS_AVAILABLE:u,IS_NODE_ENV:_,IS_FS_AVAILABLE:f,IS_PATH_AVAILABLE:b}),y=f&&b;let C="./";if(y){const K=Object({url:self.location.href}).url;K?C=o.dirname(o.dirname(n.fileURLToPath(K))):typeof __dirname<"u"&&(C=o.dirname(__dirname))}const x=y?o.join(C,"/.cache/"):null,M="/models/",T=y?o.join(C,M):M,v={version:a,backends:{onnx:{}},allowRemoteModels:!0,remoteHost:"https://huggingface.co/",remotePathTemplate:"{model}/resolve/{revision}/",allowLocalModels:!(i||l),localModelPath:T,useFS:f,useBrowserCache:c&&!A,useFSCache:f,cacheDir:x,useCustomCache:!1,customCache:null};function P(K){return Object.keys(K).length===0}},"./src/generation/configuration_utils.js":(e,r,t)=>{t.r(r),t.d(r,{GenerationConfig:()=>o});var s=t("./src/utils/core.js");class o{constructor(a){re(this,"max_length",20);re(this,"max_new_tokens",null);re(this,"min_length",0);re(this,"min_new_tokens",null);re(this,"early_stopping",!1);re(this,"max_time",null);re(this,"do_sample",!1);re(this,"num_beams",1);re(this,"num_beam_groups",1);re(this,"penalty_alpha",null);re(this,"use_cache",!0);re(this,"temperature",1);re(this,"top_k",50);re(this,"top_p",1);re(this,"typical_p",1);re(this,"epsilon_cutoff",0);re(this,"eta_cutoff",0);re(this,"diversity_penalty",0);re(this,"repetition_penalty",1);re(this,"encoder_repetition_penalty",1);re(this,"length_penalty",1);re(this,"no_repeat_ngram_size",0);re(this,"bad_words_ids",null);re(this,"force_words_ids",null);re(this,"renormalize_logits",!1);re(this,"constraints",null);re(this,"forced_bos_token_id",null);re(this,"forced_eos_token_id",null);re(this,"remove_invalid_values",!1);re(this,"exponential_decay_length_penalty",null);re(this,"suppress_tokens",null);re(this,"streamer",null);re(this,"begin_suppress_tokens",null);re(this,"forced_decoder_ids",null);re(this,"guidance_scale",null);re(this,"num_return_sequences",1);re(this,"output_attentions",!1);re(this,"output_hidden_states",!1);re(this,"output_scores",!1);re(this,"return_dict_in_generate",!1);re(this,"pad_token_id",null);re(this,"bos_token_id",null);re(this,"eos_token_id",null);re(this,"encoder_no_repeat_ngram_size",0);re(this,"decoder_start_token_id",null);re(this,"generation_kwargs",{});Object.assign(this,(0,s.pick)(a,Object.getOwnPropertyNames(this)))}}},"./src/generation/logits_process.js":(e,r,t)=>{t.r(r),t.d(r,{ClassifierFreeGuidanceLogitsProcessor:()=>g,ForcedBOSTokenLogitsProcessor:()=>l,ForcedEOSTokenLogitsProcessor:()=>c,LogitsProcessor:()=>n,LogitsProcessorList:()=>i,LogitsWarper:()=>a,MinLengthLogitsProcessor:()=>f,MinNewTokensLengthLogitsProcessor:()=>b,NoBadWordsLogitsProcessor:()=>A,NoRepeatNGramLogitsProcessor:()=>u,RepetitionPenaltyLogitsProcessor:()=>_,SuppressTokensAtBeginLogitsProcessor:()=>p,TemperatureLogitsWarper:()=>y,TopKLogitsWarper:()=>x,TopPLogitsWarper:()=>C,WhisperTimeStampLogitsProcessor:()=>d});var s=t("./src/utils/generic.js");t("./src/utils/tensor.js");var o=t("./src/utils/maths.js");class n extends s.Callable{_call(T,v){throw Error("`_call` should be implemented in a subclass")}}class a extends s.Callable{_call(T,v){throw Error("`_call` should be implemented in a subclass")}}class i extends s.Callable{constructor(){super(),this.processors=[]}push(T){this.processors.push(T)}extend(T){this.processors.push(...T)}_call(T,v){let P=v;for(const F of this.processors)P=F(T,P);return P}[Symbol.iterator](){return this.processors.values()}}class l extends n{constructor(T){super(),this.bos_token_id=T}_call(T,v){for(let P=0;P<T.length;++P)if(T[P].length===1){const F=v[P].data;F.fill(-1/0),F[this.bos_token_id]=0}return v}}class c extends n{constructor(T,v){super(),this.max_length=T,this.eos_token_id=Array.isArray(v)?v:[v]}_call(T,v){for(let P=0;P<T.length;++P)if(T[P].length===this.max_length-1){const F=v[P].data;F.fill(-1/0);for(const D of this.eos_token_id)F[D]=0}return v}}class p extends n{constructor(T,v){super(),this.begin_suppress_tokens=T,this.begin_index=v}_call(T,v){for(let P=0;P<T.length;++P)if(T[P].length===this.begin_index){const F=v[P].data;for(const D of this.begin_suppress_tokens)F[D]=-1/0}return v}}class d extends n{constructor(T,v){super(),this.eos_token_id=Array.isArray(T.eos_token_id)?T.eos_token_id[0]:T.eos_token_id,this.no_timestamps_token_id=T.no_timestamps_token_id,this.timestamp_begin=this.no_timestamps_token_id+1,this.begin_index=v.length,v.at(-1)===this.no_timestamps_token_id&&(this.begin_index-=1),this.max_initial_timestamp_index=T.max_initial_timestamp_index}_call(T,v){for(let P=0;P<T.length;++P){const F=v[P].data;if(F[this.no_timestamps_token_id]=-1/0,T[P].length===this.begin_index-1){F.fill(-1/0),F[this.timestamp_begin]=0;continue}const D=T[P].slice(this.begin_index),K=D.length>=1&&D[D.length-1]>=this.timestamp_begin,U=D.length<2||D[D.length-2]>=this.timestamp_begin;if(K&&(U?F.subarray(this.timestamp_begin).fill(-1/0):F.subarray(0,this.eos_token_id).fill(-1/0)),T[P].length===this.begin_index&&this.max_initial_timestamp_index!==null){const te=this.timestamp_begin+this.max_initial_timestamp_index;F.subarray(te+1).fill(-1/0)}const j=(0,o.log_softmax)(F),ne=Math.log(j.subarray(this.timestamp_begin).map(Math.exp).reduce((te,Z)=>te+Z)),q=(0,o.max)(j.subarray(0,this.timestamp_begin))[0];ne>q&&F.subarray(0,this.timestamp_begin).fill(-1/0)}return v}}class u extends n{constructor(T){super(),this.no_repeat_ngram_size=T}getNgrams(T){const v=T.length,P=[];for(let D=0;D<v+1-this.no_repeat_ngram_size;++D){const K=[];for(let U=0;U<this.no_repeat_ngram_size;++U)K.push(T[D+U]);P.push(K.map(Number))}const F=new Map;for(const D of P){const K=D.slice(0,D.length-1),U=JSON.stringify(K),j=F.get(U)??[];j.push(D[D.length-1]),F.set(U,j)}return F}getGeneratedNgrams(T,v){const P=v.slice(v.length+1-this.no_repeat_ngram_size,v.length);return T.get(JSON.stringify(P.map(Number)))??[]}calcBannedNgramTokens(T){const v=[];if(T.length+1<this.no_repeat_ngram_size)return v;{const P=this.getNgrams(T);return this.getGeneratedNgrams(P,T)}}_call(T,v){for(let P=0;P<T.length;++P){const F=v[P].data,D=this.calcBannedNgramTokens(T[P]);for(const K of D)F[K]=-1/0}return v}}class _ extends n{constructor(T){super(),this.penalty=T}_call(T,v){for(let P=0;P<T.length;++P){const F=v[P].data;for(const D of new Set(T[P])){const K=Number(D);F[K]<0?F[K]*=this.penalty:F[K]/=this.penalty}}return v}}class f extends n{constructor(T,v){super(),this.min_length=T,this.eos_token_id=Array.isArray(v)?v:[v]}_call(T,v){for(let P=0;P<T.length;++P)if(T[P].length<this.min_length){const F=v[P].data;for(const D of this.eos_token_id)F[D]=-1/0}return v}}class b extends n{constructor(T,v,P){super(),this.prompt_length_to_skip=T,this.min_new_tokens=v,this.eos_token_id=Array.isArray(P)?P:[P]}_call(T,v){for(let P=0;P<T.length;++P)if(T[P].length-this.prompt_length_to_skip<this.min_new_tokens){const D=v[P].data;for(const K of this.eos_token_id)D[K]=-1/0}return v}}class A extends n{constructor(T,v){super(),this.bad_words_ids=T,this.eos_token_id=Array.isArray(v)?v:[v]}_call(T,v){for(let P=0;P<T.length;++P){const F=v[P].data,D=T[P];for(const K of this.bad_words_ids){if(D.length<K.length-1)continue;let U=!0;for(let j=1;j<=K.length-1;++j)if(K.at(-j-1)!=D.at(-j)){U=!1;break}U&&(F[K.at(-1)]=-1/0)}}return v}}class g extends n{constructor(T){if(super(),T<=1)throw new Error(`Require guidance scale >1 to use the classifier free guidance processor, got guidance scale ${T}.`);this.guidance_scale=T}_call(T,v){if(v.dims[0]!==2*T.length)throw new Error(`Logits should have twice the batch size of the input ids, the first half of batches corresponding to the conditional inputs, and the second half of batches corresponding to the unconditional inputs. Got batch size ${v.dims[0]} for the logits and ${T.length} for the input ids.`);const P=T.length,F=v.slice([0,P],null),D=v.slice([P,v.dims[0]],null);for(let K=0;K<D.data.length;++K)D.data[K]+=(F.data[K]-D.data[K])*this.guidance_scale;return D}}class y extends a{constructor(T){super(),this.temperature=T}_call(T,v){const P=v.data;for(let F=0;F<P.length;++F)P[F]/=this.temperature;return v}}class C extends a{constructor(T,{filter_value:v=-1/0,min_tokens_to_keep:P=1}={}){if(super(),T<0||T>1)throw new Error(`\`top_p\` must be a float > 0 and < 1, but is ${T}`);if(!Number.isInteger(P)||P<1)throw new Error(`\`min_tokens_to_keep\` must be a positive integer, but is ${P}`);this.top_p=T,this.filter_value=v,this.min_tokens_to_keep=P}}class x extends a{constructor(T,{filter_value:v=-1/0,min_tokens_to_keep:P=1}={}){if(super(),!Number.isInteger(T)||T<0)throw new Error(`\`top_k\` must be a positive integer, but is ${T}`);this.top_k=Math.max(T,P),this.filter_value=v}}},"./src/generation/logits_sampler.js":(e,r,t)=>{t.r(r),t.d(r,{LogitsSampler:()=>a});var s=t("./src/utils/generic.js"),o=t("./src/utils/tensor.js"),n=t("./src/utils/maths.js");t("./src/generation/configuration_utils.js");class a extends s.Callable{constructor(d){super(),this.generation_config=d}async _call(d){return this.sample(d)}async sample(d){throw Error("sample should be implemented in subclasses.")}getLogits(d,u){let _=d.dims.at(-1),f=d.data;if(u===-1)f=f.slice(-_);else{let b=u*_;f=f.slice(b,b+_)}return f}randomSelect(d){let u=0;for(let f=0;f<d.length;++f)u+=d[f];let _=Math.random()*u;for(let f=0;f<d.length;++f)if(_-=d[f],_<=0)return f;return 0}static getSampler(d){if(d.do_sample)return new l(d);if(d.num_beams>1)return new c(d);if(d.num_return_sequences>1)throw Error(`num_return_sequences has to be 1 when doing greedy search, but is ${d.num_return_sequences}.`);return new i(d)}}class i extends a{async sample(d){const u=(0,n.max)(d.data)[1];return[[BigInt(u),0]]}}class l extends a{async sample(d){let u=d.dims.at(-1);this.generation_config.top_k>0&&(u=Math.min(this.generation_config.top_k,u));const[_,f]=await(0,o.topk)(d,u),b=(0,n.softmax)(_.data);return Array.from({length:this.generation_config.num_beams},()=>{const A=this.randomSelect(b);return[f.data[A],Math.log(b[A])]})}}class c extends a{async sample(d){let u=d.dims.at(-1);this.generation_config.top_k>0&&(u=Math.min(this.generation_config.top_k,u));const[_,f]=await(0,o.topk)(d,u),b=(0,n.softmax)(_.data);return Array.from({length:this.generation_config.num_beams},(A,g)=>[f.data[g],Math.log(b[g])])}}},"./src/generation/stopping_criteria.js":(e,r,t)=>{t.r(r),t.d(r,{EosTokenCriteria:()=>i,InterruptableStoppingCriteria:()=>l,MaxLengthCriteria:()=>a,StoppingCriteria:()=>o,StoppingCriteriaList:()=>n});var s=t("./src/utils/generic.js");class o extends s.Callable{_call(p,d){throw Error("StoppingCriteria needs to be subclassed")}}class n extends s.Callable{constructor(){super(),this.criteria=[]}push(p){this.criteria.push(p)}extend(p){p instanceof n?p=p.criteria:p instanceof o&&(p=[p]),this.criteria.push(...p)}_call(p,d){const u=new Array(p.length).fill(!1);for(const _ of this.criteria){const f=_(p,d);for(let b=0;b<u.length;++b)u[b]||(u[b]=f[b])}return u}[Symbol.iterator](){return this.criteria.values()}}class a extends o{constructor(p,d=null){super(),this.max_length=p,this.max_position_embeddings=d}_call(p){return p.map(d=>d.length>=this.max_length)}}class i extends o{constructor(p){super(),Array.isArray(p)||(p=[p]),this.eos_token_id=p}_call(p,d){return p.map(u=>{const _=u.at(-1);return this.eos_token_id.some(f=>_==f)})}}class l extends o{constructor(){super(),this.interrupted=!1}interrupt(){this.interrupted=!0}reset(){this.interrupted=!1}_call(p,d){return new Array(p.length).fill(this.interrupted)}}},"./src/generation/streamers.js":(e,r,t)=>{t.r(r),t.d(r,{BaseStreamer:()=>a,TextStreamer:()=>l,WhisperTextStreamer:()=>c});var s=t("./src/utils/core.js"),o=t("./src/tokenizers.js"),n=t("./src/env.js");class a{put(d){throw Error("Not implemented")}end(){throw Error("Not implemented")}}const i=n.apis.IS_PROCESS_AVAILABLE?p=>process.stdout.write(p):p=>console.log(p);class l extends a{constructor(d,{skip_prompt:u=!1,callback_function:_=null,token_callback_function:f=null,skip_special_tokens:b=!0,decode_kwargs:A={},...g}={}){super(),this.tokenizer=d,this.skip_prompt=u,this.callback_function=_??i,this.token_callback_function=f,this.decode_kwargs={skip_special_tokens:b,...A,...g},this.token_cache=[],this.print_len=0,this.next_tokens_are_prompt=!0}put(d){var A;if(d.length>1)throw Error("TextStreamer only supports batch size of 1");const u=this.next_tokens_are_prompt;if(u&&(this.next_tokens_are_prompt=!1,this.skip_prompt))return;const _=d[0];(A=this.token_callback_function)==null||A.call(this,_),this.token_cache=(0,s.mergeArrays)(this.token_cache,_);const f=this.tokenizer.decode(this.token_cache,this.decode_kwargs);let b;u||f.endsWith(`
`)?(b=f.slice(this.print_len),this.token_cache=[],this.print_len=0):f.length>0&&(0,o.is_chinese_char)(f.charCodeAt(f.length-1))?(b=f.slice(this.print_len),this.print_len+=b.length):(b=f.slice(this.print_len,f.lastIndexOf(" ")+1),this.print_len+=b.length),this.on_finalized_text(b,!1)}end(){let d;this.token_cache.length>0?(d=this.tokenizer.decode(this.token_cache,this.decode_kwargs).slice(this.print_len),this.token_cache=[],this.print_len=0):d="",this.next_tokens_are_prompt=!0,this.on_finalized_text(d,!0)}on_finalized_text(d,u){var _,f;d.length>0&&((_=this.callback_function)==null||_.call(this,d)),u&&this.callback_function===i&&n.apis.IS_PROCESS_AVAILABLE&&((f=this.callback_function)==null||f.call(this,`
`))}}class c extends l{constructor(d,{skip_prompt:u=!1,callback_function:_=null,token_callback_function:f=null,on_chunk_start:b=null,on_chunk_end:A=null,on_finalize:g=null,time_precision:y=.02,skip_special_tokens:C=!0,decode_kwargs:x={}}={}){super(d,{skip_prompt:u,skip_special_tokens:C,callback_function:_,token_callback_function:f,decode_kwargs:x}),this.timestamp_begin=d.timestamp_begin,this.on_chunk_start=b,this.on_chunk_end=A,this.on_finalize=g,this.time_precision=y,this.waiting_for_timestamp=!1}put(d){var _,f,b;if(d.length>1)throw Error("WhisperTextStreamer only supports batch size of 1");const u=d[0];if(u.length===1){const A=Number(u[0])-this.timestamp_begin;if(A>=0){const g=A*this.time_precision;this.waiting_for_timestamp?(_=this.on_chunk_end)==null||_.call(this,g):(f=this.on_chunk_start)==null||f.call(this,g),this.waiting_for_timestamp=!this.waiting_for_timestamp,(b=this.token_callback_function)==null||b.call(this,u);return}}return super.put(d)}end(){var d;super.end(),(d=this.on_finalize)==null||d.call(this)}}},"./src/models.js":(e,r,t)=>{t.r(r),t.d(r,{ASTForAudioClassification:()=>Na,ASTModel:()=>Ba,ASTPreTrainedModel:()=>Qn,AlbertForMaskedLM:()=>et,AlbertForQuestionAnswering:()=>Qe,AlbertForSequenceClassification:()=>Xe,AlbertModel:()=>Ge,AlbertPreTrainedModel:()=>Se,ArceeForCausalLM:()=>Qs,ArceeModel:()=>Tr,ArceePreTrainedModel:()=>Ut,AutoModel:()=>Hc,AutoModelForAudioClassification:()=>Uv,AutoModelForAudioFrameClassification:()=>Gv,AutoModelForAudioTextToText:()=>ex,AutoModelForCTC:()=>Vv,AutoModelForCausalLM:()=>kv,AutoModelForDepthEstimation:()=>Qv,AutoModelForDocumentQuestionAnswering:()=>Hv,AutoModelForImageClassification:()=>Dv,AutoModelForImageFeatureExtraction:()=>Yv,AutoModelForImageMatting:()=>Kv,AutoModelForImageSegmentation:()=>Lv,AutoModelForImageTextToText:()=>Zv,AutoModelForImageToImage:()=>qv,AutoModelForMaskGeneration:()=>jv,AutoModelForMaskedLM:()=>Iv,AutoModelForNormalEstimation:()=>Xv,AutoModelForObjectDetection:()=>Bv,AutoModelForPoseEstimation:()=>Jv,AutoModelForQuestionAnswering:()=>Fv,AutoModelForSemanticSegmentation:()=>zv,AutoModelForSeq2SeqLM:()=>Cv,AutoModelForSequenceClassification:()=>Ev,AutoModelForSpeechSeq2Seq:()=>Sv,AutoModelForTextToSpectrogram:()=>$v,AutoModelForTextToWaveform:()=>Av,AutoModelForTokenClassification:()=>Pv,AutoModelForUniversalSegmentation:()=>Rv,AutoModelForVision2Seq:()=>Ov,AutoModelForXVector:()=>Wv,AutoModelForZeroShotObjectDetection:()=>Nv,BartForConditionalGeneration:()=>Yt,BartForSequenceClassification:()=>Mr,BartModel:()=>ns,BartPretrainedModel:()=>xr,BaseModelOutput:()=>Ee,BeitForImageClassification:()=>ZM,BeitModel:()=>YM,BeitPreTrainedModel:()=>ic,BertForMaskedLM:()=>ye,BertForQuestionAnswering:()=>ce,BertForSequenceClassification:()=>de,BertForTokenClassification:()=>we,BertModel:()=>Ce,BertPreTrainedModel:()=>Me,BlenderbotForConditionalGeneration:()=>fr,BlenderbotModel:()=>pr,BlenderbotPreTrainedModel:()=>ir,BlenderbotSmallForConditionalGeneration:()=>Ks,BlenderbotSmallModel:()=>Qr,BlenderbotSmallPreTrainedModel:()=>er,BloomForCausalLM:()=>yM,BloomModel:()=>bM,BloomPreTrainedModel:()=>Qu,CLIPModel:()=>Zn,CLIPPreTrainedModel:()=>As,CLIPSegForImageSegmentation:()=>In,CLIPSegModel:()=>os,CLIPSegPreTrainedModel:()=>ln,CLIPTextModel:()=>Ya,CLIPTextModelWithProjection:()=>zo,CLIPVisionModel:()=>Za,CLIPVisionModelWithProjection:()=>Ro,CamembertForMaskedLM:()=>wt,CamembertForQuestionAnswering:()=>tt,CamembertForSequenceClassification:()=>pt,CamembertForTokenClassification:()=>xt,CamembertModel:()=>nt,CamembertPreTrainedModel:()=>De,CausalLMOutput:()=>cn,CausalLMOutputWithPast:()=>IE,ChineseCLIPModel:()=>kn,ChineseCLIPPreTrainedModel:()=>jo,ClapAudioModelWithProjection:()=>i0,ClapModel:()=>o0,ClapPreTrainedModel:()=>li,ClapTextModelWithProjection:()=>a0,CodeGenForCausalLM:()=>R,CodeGenModel:()=>I,CodeGenPreTrainedModel:()=>E,CohereForCausalLM:()=>Zw,CohereModel:()=>Yw,CoherePreTrainedModel:()=>Bu,ConvBertForMaskedLM:()=>Os,ConvBertForQuestionAnswering:()=>Kt,ConvBertForSequenceClassification:()=>Ds,ConvBertForTokenClassification:()=>St,ConvBertModel:()=>ft,ConvBertPreTrainedModel:()=>ds,ConvNextForImageClassification:()=>Kb,ConvNextModel:()=>Hb,ConvNextPreTrainedModel:()=>vc,ConvNextV2ForImageClassification:()=>Qb,ConvNextV2Model:()=>qb,ConvNextV2PreTrainedModel:()=>xc,DFineForObjectDetection:()=>pb,DFineModel:()=>db,DFinePreTrainedModel:()=>hc,DINOv3ConvNextModel:()=>sy,DINOv3ConvNextPreTrainedModel:()=>ry,DINOv3ViTModel:()=>ty,DINOv3ViTPreTrainedModel:()=>ey,DPTForDepthEstimation:()=>Sb,DPTModel:()=>Cb,DPTPreTrainedModel:()=>Mc,DacDecoderModel:()=>Q0,DacDecoderOutput:()=>H0,DacEncoderModel:()=>q0,DacEncoderOutput:()=>G0,DacModel:()=>K0,DacPreTrainedModel:()=>fi,DebertaForMaskedLM:()=>Wr,DebertaForQuestionAnswering:()=>kr,DebertaForSequenceClassification:()=>qr,DebertaForTokenClassification:()=>nr,DebertaModel:()=>qt,DebertaPreTrainedModel:()=>It,DebertaV2ForMaskedLM:()=>hs,DebertaV2ForQuestionAnswering:()=>zs,DebertaV2ForSequenceClassification:()=>Ir,DebertaV2ForTokenClassification:()=>Ls,DebertaV2Model:()=>ps,DebertaV2PreTrainedModel:()=>cr,DecisionTransformerModel:()=>S0,DecisionTransformerPreTrainedModel:()=>C0,DeiTForImageClassification:()=>gb,DeiTModel:()=>_b,DeiTPreTrainedModel:()=>fc,DepthAnythingForDepthEstimation:()=>Ab,DepthAnythingPreTrainedModel:()=>$b,DepthProForDepthEstimation:()=>Db,DepthProPreTrainedModel:()=>Ob,DetrForObjectDetection:()=>tb,DetrForSegmentation:()=>lc,DetrModel:()=>eb,DetrObjectDetectionOutput:()=>uc,DetrPreTrainedModel:()=>ri,DetrSegmentationOutput:()=>rb,Dinov2ForImageClassification:()=>Jb,Dinov2Model:()=>Xb,Dinov2PreTrainedModel:()=>Tc,Dinov2WithRegistersForImageClassification:()=>Zb,Dinov2WithRegistersModel:()=>Yb,Dinov2WithRegistersPreTrainedModel:()=>Ec,DistilBertForMaskedLM:()=>mr,DistilBertForQuestionAnswering:()=>Cs,DistilBertForSequenceClassification:()=>Yr,DistilBertForTokenClassification:()=>Rr,DistilBertModel:()=>ms,DistilBertPreTrainedModel:()=>vr,DonutSwinModel:()=>Gb,DonutSwinPreTrainedModel:()=>Wb,EfficientNetForImageClassification:()=>m0,EfficientNetModel:()=>h0,EfficientNetPreTrainedModel:()=>Lc,ElectraForMaskedLM:()=>V,ElectraForQuestionAnswering:()=>xe,ElectraForSequenceClassification:()=>Y,ElectraForTokenClassification:()=>oe,ElectraModel:()=>ee,ElectraPreTrainedModel:()=>$,Ernie4_5_ForCausalLM:()=>e0,Ernie4_5_Model:()=>Zy,Ernie4_5_PretrainedModel:()=>kc,EsmForMaskedLM:()=>Gs,EsmForSequenceClassification:()=>Gr,EsmForTokenClassification:()=>Ne,EsmModel:()=>fs,EsmPreTrainedModel:()=>ar,ExaoneForCausalLM:()=>jw,ExaoneModel:()=>Nw,ExaonePreTrainedModel:()=>Fu,FalconForCausalLM:()=>n0,FalconModel:()=>s0,FalconPreTrainedModel:()=>Fc,FastViTForImageClassification:()=>jM,FastViTModel:()=>NM,FastViTPreTrainedModel:()=>rc,Florence2ForConditionalGeneration:()=>Ha,Florence2PreTrainedModel:()=>Ga,GLPNForDepthEstimation:()=>Ub,GLPNModel:()=>Vb,GLPNPreTrainedModel:()=>yc,GPT2LMHeadModel:()=>Go,GPT2Model:()=>Wo,GPT2PreTrainedModel:()=>Fn,GPTBigCodeForCausalLM:()=>h,GPTBigCodeModel:()=>Jo,GPTBigCodePreTrainedModel:()=>ao,GPTJForCausalLM:()=>Xo,GPTJModel:()=>Qo,GPTJPreTrainedModel:()=>oo,GPTNeoForCausalLM:()=>On,GPTNeoModel:()=>Ko,GPTNeoPreTrainedModel:()=>so,GPTNeoXForCausalLM:()=>qo,GPTNeoXModel:()=>no,GPTNeoXPreTrainedModel:()=>Dn,Gemma2ForCausalLM:()=>sM,Gemma2Model:()=>rM,Gemma2PreTrainedModel:()=>ju,Gemma3ForCausalLM:()=>iM,Gemma3Model:()=>aM,Gemma3PreTrainedModel:()=>Uu,Gemma3nForConditionalGeneration:()=>Do,Gemma3nPreTrainedModel:()=>Xa,GemmaForCausalLM:()=>tM,GemmaModel:()=>eM,GemmaPreTrainedModel:()=>Nu,GlmForCausalLM:()=>Bw,GlmModel:()=>Rw,GlmPreTrainedModel:()=>Iu,GraniteForCausalLM:()=>Qw,GraniteModel:()=>qw,GraniteMoeHybridForCausalLM:()=>Jw,GraniteMoeHybridModel:()=>Xw,GraniteMoeHybridPreTrainedModel:()=>Ru,GranitePreTrainedModel:()=>zu,GroundingDinoForObjectDetection:()=>oy,GroundingDinoPreTrainedModel:()=>ny,GroupViTModel:()=>BM,GroupViTPreTrainedModel:()=>RM,HeliumForCausalLM:()=>zw,HeliumModel:()=>Lw,HeliumPreTrainedModel:()=>ku,HieraForImageClassification:()=>Mb,HieraModel:()=>wb,HieraPreTrainedModel:()=>_c,HubertForCTC:()=>zy,HubertForSequenceClassification:()=>Ry,HubertModel:()=>Ly,HubertPreTrainedModel:()=>wE,IJepaForImageClassification:()=>$M,IJepaModel:()=>SM,IJepaPreTrainedModel:()=>Zu,Idefics3ForConditionalGeneration:()=>Bs,Idefics3PreTrainedModel:()=>An,ImageMattingOutput:()=>rx,JAISLMHeadModel:()=>Ho,JAISModel:()=>ro,JAISPreTrainedModel:()=>to,JinaCLIPModel:()=>ti,JinaCLIPPreTrainedModel:()=>ut,JinaCLIPTextModel:()=>Vo,JinaCLIPVisionModel:()=>Uo,Lfm2ForCausalLM:()=>Fw,Lfm2Model:()=>Iw,Lfm2PreTrainedModel:()=>Yo,LiteWhisperForConditionalGeneration:()=>ja,Llama4ForCausalLM:()=>Je,Llama4PreTrainedModel:()=>Re,LlamaForCausalLM:()=>Ie,LlamaModel:()=>pe,LlamaPreTrainedModel:()=>H,LlavaForConditionalGeneration:()=>$n,LlavaOnevisionForConditionalGeneration:()=>Ua,LlavaPreTrainedModel:()=>Sn,LlavaQwen2ForCausalLM:()=>Qa,LongT5ForConditionalGeneration:()=>_r,LongT5Model:()=>Gt,LongT5PreTrainedModel:()=>Zt,M2M100ForConditionalGeneration:()=>fy,M2M100Model:()=>my,M2M100PreTrainedModel:()=>Sc,MBartForCausalLM:()=>_s,MBartForConditionalGeneration:()=>$s,MBartForSequenceClassification:()=>Or,MBartModel:()=>Zr,MBartPreTrainedModel:()=>Fr,MPNetForMaskedLM:()=>En,MPNetForQuestionAnswering:()=>ge,MPNetForSequenceClassification:()=>Pn,MPNetForTokenClassification:()=>Cn,MPNetModel:()=>Tn,MPNetPreTrainedModel:()=>ss,MT5ForConditionalGeneration:()=>wr,MT5Model:()=>dr,MT5PreTrainedModel:()=>gr,MarianMTModel:()=>hy,MarianModel:()=>py,MarianPreTrainedModel:()=>Cc,MaskFormerForInstanceSegmentation:()=>jb,MaskFormerModel:()=>Nb,MaskFormerPreTrainedModel:()=>bc,MaskedLMOutput:()=>Dr,Metric3DForDepthEstimation:()=>zb,Metric3DPreTrainedModel:()=>Lb,Metric3Dv2ForDepthEstimation:()=>Bb,Metric3Dv2PreTrainedModel:()=>Rb,MgpstrForSceneTextRecognition:()=>F0,MgpstrModelOutput:()=>k0,MgpstrPreTrainedModel:()=>I0,MimiDecoderModel:()=>W0,MimiDecoderOutput:()=>j0,MimiEncoderModel:()=>U0,MimiEncoderOutput:()=>N0,MimiModel:()=>V0,MimiPreTrainedModel:()=>mi,MistralForCausalLM:()=>Yy,MistralModel:()=>Jy,MistralPreTrainedModel:()=>Ac,MobileBertForMaskedLM:()=>Qt,MobileBertForQuestionAnswering:()=>Ss,MobileBertForSequenceClassification:()=>Hs,MobileBertModel:()=>rt,MobileBertPreTrainedModel:()=>je,MobileLLMForCausalLM:()=>Uw,MobileLLMModel:()=>Vw,MobileLLMPreTrainedModel:()=>Ou,MobileNetV1ForImageClassification:()=>_0,MobileNetV1ForSemanticSegmentation:()=>g0,MobileNetV1Model:()=>f0,MobileNetV1PreTrainedModel:()=>ci,MobileNetV2ForImageClassification:()=>M0,MobileNetV2ForSemanticSegmentation:()=>b0,MobileNetV2Model:()=>w0,MobileNetV2PreTrainedModel:()=>di,MobileNetV3ForImageClassification:()=>v0,MobileNetV3ForSemanticSegmentation:()=>x0,MobileNetV3Model:()=>y0,MobileNetV3PreTrainedModel:()=>pi,MobileNetV4ForImageClassification:()=>E0,MobileNetV4ForSemanticSegmentation:()=>P0,MobileNetV4Model:()=>T0,MobileNetV4PreTrainedModel:()=>hi,MobileViTForImageClassification:()=>GM,MobileViTModel:()=>WM,MobileViTPreTrainedModel:()=>sc,MobileViTV2ForImageClassification:()=>KM,MobileViTV2Model:()=>HM,MobileViTV2PreTrainedModel:()=>nc,ModelOutput:()=>_e,ModernBertDecoderForCausalLM:()=>yr,ModernBertDecoderModel:()=>ht,ModernBertDecoderPreTrainedModel:()=>kt,ModernBertForMaskedLM:()=>He,ModernBertForSequenceClassification:()=>gt,ModernBertForTokenClassification:()=>dt,ModernBertModel:()=>ze,ModernBertPreTrainedModel:()=>Ze,Moondream1ForConditionalGeneration:()=>Wa,MoonshineForConditionalGeneration:()=>Va,MoonshineModel:()=>$u,MoonshinePreTrainedModel:()=>Jn,MptForCausalLM:()=>xM,MptModel:()=>vM,MptPreTrainedModel:()=>Xu,MultiModalityCausalLM:()=>A0,MultiModalityPreTrainedModel:()=>$0,MusicgenForCausalLM:()=>vE,MusicgenForConditionalGeneration:()=>Rc,MusicgenModel:()=>yE,MusicgenPreTrainedModel:()=>zc,NanoChatForCausalLM:()=>zt,NanoChatModel:()=>Tt,NanoChatPreTrainedModel:()=>ot,NeoBertForMaskedLM:()=>Te,NeoBertForQuestionAnswering:()=>st,NeoBertForSequenceClassification:()=>We,NeoBertForTokenClassification:()=>qe,NeoBertModel:()=>Le,NeoBertPreTrainedModel:()=>ke,NomicBertModel:()=>Vr,NomicBertPreTrainedModel:()=>$r,OPTForCausalLM:()=>EM,OPTModel:()=>TM,OPTPreTrainedModel:()=>Ju,Olmo2ForCausalLM:()=>Kw,Olmo2Model:()=>Hw,Olmo2PreTrainedModel:()=>Lu,OlmoForCausalLM:()=>Gw,OlmoModel:()=>Ww,OlmoPreTrainedModel:()=>Du,OpenELMForCausalLM:()=>uM,OpenELMModel:()=>lM,OpenELMPreTrainedModel:()=>Wu,OwlViTForObjectDetection:()=>QM,OwlViTModel:()=>qM,OwlViTPreTrainedModel:()=>oc,Owlv2ForObjectDetection:()=>JM,Owlv2Model:()=>XM,Owlv2PreTrainedModel:()=>ac,PaliGemmaForConditionalGeneration:()=>qa,PaliGemmaPreTrainedModel:()=>Ka,ParakeetForCTC:()=>yy,ParakeetPreTrainedModel:()=>by,PatchTSMixerForPrediction:()=>z0,PatchTSMixerModel:()=>L0,PatchTSMixerPreTrainedModel:()=>Nc,PatchTSTForPrediction:()=>D0,PatchTSTModel:()=>O0,PatchTSTPreTrainedModel:()=>Bc,Phi3ForCausalLM:()=>MM,Phi3Model:()=>wM,Phi3PreTrainedModel:()=>qu,Phi3VForCausalLM:()=>Lo,Phi3VPreTrainedModel:()=>Ja,PhiForCausalLM:()=>gM,PhiModel:()=>_M,PhiPreTrainedModel:()=>Ku,PreTrainedModel:()=>z,PretrainedMixin:()=>Nt,PvtForImageClassification:()=>FM,PvtModel:()=>IM,PvtPreTrainedModel:()=>ec,PyAnnoteForAudioFrameClassification:()=>xy,PyAnnoteModel:()=>vy,PyAnnotePreTrainedModel:()=>$c,QuestionAnsweringModelOutput:()=>Nr,Qwen2ForCausalLM:()=>dM,Qwen2Model:()=>cM,Qwen2PreTrainedModel:()=>Gu,Qwen2VLForConditionalGeneration:()=>fM,Qwen2VLPreTrainedModel:()=>mM,Qwen3ForCausalLM:()=>hM,Qwen3Model:()=>pM,Qwen3PreTrainedModel:()=>Hu,RFDetrForObjectDetection:()=>ub,RFDetrModel:()=>lb,RFDetrObjectDetectionOutput:()=>cb,RFDetrPreTrainedModel:()=>pc,RTDetrForObjectDetection:()=>nb,RTDetrModel:()=>sb,RTDetrObjectDetectionOutput:()=>Zo,RTDetrPreTrainedModel:()=>cc,RTDetrV2ForObjectDetection:()=>ab,RTDetrV2Model:()=>ob,RTDetrV2ObjectDetectionOutput:()=>ib,RTDetrV2PreTrainedModel:()=>dc,ResNetForImageClassification:()=>yb,ResNetModel:()=>bb,ResNetPreTrainedModel:()=>gc,RoFormerForMaskedLM:()=>Ar,RoFormerForQuestionAnswering:()=>nn,RoFormerForSequenceClassification:()=>rn,RoFormerForTokenClassification:()=>sn,RoFormerModel:()=>sr,RoFormerPreTrainedModel:()=>Ur,RobertaForMaskedLM:()=>Sa,RobertaForQuestionAnswering:()=>Aa,RobertaForSequenceClassification:()=>Ao,RobertaForTokenClassification:()=>$a,RobertaModel:()=>Ca,RobertaPreTrainedModel:()=>Rs,SamImageSegmentationOutput:()=>dy,SamModel:()=>cy,SamPreTrainedModel:()=>uy,SapiensForDepthEstimation:()=>Ib,SapiensForNormalEstimation:()=>Fb,SapiensForSemanticSegmentation:()=>kb,SapiensPreTrainedModel:()=>ni,SegformerForImageClassification:()=>u0,SegformerForSemanticSegmentation:()=>c0,SegformerModel:()=>bE,SegformerPreTrainedModel:()=>ui,Seq2SeqLMOutput:()=>kE,SequenceClassifierOutput:()=>Et,SiglipModel:()=>Bo,SiglipPreTrainedModel:()=>eo,SiglipTextModel:()=>ei,SiglipVisionModel:()=>No,SmolLM3ForCausalLM:()=>Dw,SmolLM3Model:()=>Ow,SmolLM3PreTrainedModel:()=>Au,SmolVLMForConditionalGeneration:()=>Yn,SnacDecoderModel:()=>Y0,SnacEncoderModel:()=>J0,SnacModel:()=>X0,SnacPreTrainedModel:()=>_i,SpeechT5ForSpeechToText:()=>Hy,SpeechT5ForTextToSpeech:()=>Ky,SpeechT5HifiGan:()=>qy,SpeechT5Model:()=>ME,SpeechT5PreTrainedModel:()=>ii,SqueezeBertForMaskedLM:()=>se,SqueezeBertForQuestionAnswering:()=>fe,SqueezeBertForSequenceClassification:()=>ue,SqueezeBertModel:()=>G,SqueezeBertPreTrainedModel:()=>k,StableLmForCausalLM:()=>p0,StableLmModel:()=>d0,StableLmPreTrainedModel:()=>Dc,Starcoder2ForCausalLM:()=>r0,Starcoder2Model:()=>t0,Starcoder2PreTrainedModel:()=>Ic,StyleTextToSpeech2Model:()=>Gy,StyleTextToSpeech2PreTrainedModel:()=>Wy,Swin2SRForImageSuperResolution:()=>Pb,Swin2SRModel:()=>Eb,Swin2SRPreTrainedModel:()=>wc,SwinForImageClassification:()=>xb,SwinForSemanticSegmentation:()=>Tb,SwinModel:()=>vb,SwinPreTrainedModel:()=>si,T5ForConditionalGeneration:()=>Lt,T5Model:()=>Rt,T5PreTrainedModel:()=>bt,TableTransformerForObjectDetection:()=>mb,TableTransformerModel:()=>hb,TableTransformerObjectDetectionOutput:()=>fb,TableTransformerPreTrainedModel:()=>mc,TokenClassifierOutput:()=>Er,TrOCRForCausalLM:()=>Xy,TrOCRPreTrainedModel:()=>Qy,UltravoxModel:()=>jc,UltravoxPreTrainedModel:()=>R0,UniSpeechForCTC:()=>Cy,UniSpeechForSequenceClassification:()=>Sy,UniSpeechModel:()=>Py,UniSpeechPreTrainedModel:()=>oi,UniSpeechSatForAudioFrameClassification:()=>Iy,UniSpeechSatForCTC:()=>Ay,UniSpeechSatForSequenceClassification:()=>ky,UniSpeechSatModel:()=>$y,UniSpeechSatPreTrainedModel:()=>ea,VaultGemmaForCausalLM:()=>oM,VaultGemmaModel:()=>nM,VaultGemmaPreTrainedModel:()=>Vu,ViTForImageClassification:()=>CM,ViTMAEModel:()=>DM,ViTMAEPreTrainedModel:()=>OM,ViTMSNForImageClassification:()=>zM,ViTMSNModel:()=>LM,ViTMSNPreTrainedModel:()=>tc,ViTModel:()=>PM,ViTPreTrainedModel:()=>Yu,VisionEncoderDecoderModel:()=>Oo,VitMatteForImageMatting:()=>UM,VitMattePreTrainedModel:()=>VM,VitPoseForPoseEstimation:()=>kM,VitPosePreTrainedModel:()=>AM,VitsModel:()=>Oc,VitsModelOutput:()=>sx,VitsPreTrainedModel:()=>l0,VoxtralForConditionalGeneration:()=>B0,Wav2Vec2BertForCTC:()=>Oy,Wav2Vec2BertForSequenceClassification:()=>Dy,Wav2Vec2BertModel:()=>Fy,Wav2Vec2BertPreTrainedModel:()=>ai,Wav2Vec2ForAudioFrameClassification:()=>My,Wav2Vec2ForCTC:()=>gy,Wav2Vec2ForSequenceClassification:()=>wy,Wav2Vec2Model:()=>_y,Wav2Vec2PreTrainedModel:()=>un,WavLMForAudioFrameClassification:()=>Uy,WavLMForCTC:()=>Ny,WavLMForSequenceClassification:()=>jy,WavLMForXVector:()=>Vy,WavLMModel:()=>By,WavLMPreTrainedModel:()=>io,WeSpeakerResNetModel:()=>Ey,WeSpeakerResNetPreTrainedModel:()=>Ty,WhisperForConditionalGeneration:()=>qs,WhisperModel:()=>Fo,WhisperPreTrainedModel:()=>Xn,XLMForQuestionAnswering:()=>Da,XLMForSequenceClassification:()=>Fa,XLMForTokenClassification:()=>Oa,XLMModel:()=>ka,XLMPreTrainedModel:()=>on,XLMRobertaForMaskedLM:()=>za,XLMRobertaForQuestionAnswering:()=>Io,XLMRobertaForSequenceClassification:()=>Ra,XLMRobertaForTokenClassification:()=>ko,XLMRobertaModel:()=>La,XLMRobertaPreTrainedModel:()=>an,XLMWithLMHeadModel:()=>Ia,XVectorOutput:()=>tx,YolosForObjectDetection:()=>iy,YolosModel:()=>ay,YolosObjectDetectionOutput:()=>ly,YolosPreTrainedModel:()=>Pc});var s=t("./src/configs.js"),o=t("./src/backends/onnx.js"),n=t("./src/utils/dtypes.js"),a=t("./src/utils/generic.js"),i=t("./src/utils/core.js"),l=t("./src/utils/hub.js"),c=t("./src/utils/constants.js"),p=t("./src/generation/logits_process.js"),d=t("./src/generation/configuration_utils.js"),u=t("./src/utils/tensor.js"),_=t("./src/utils/image.js"),f=t("./src/utils/maths.js"),b=t("./src/generation/stopping_criteria.js"),A=t("./src/generation/logits_sampler.js"),g=t("./src/env.js"),y=t("./src/models/whisper/generation_whisper.js"),C=t("./src/models/whisper/common_whisper.js");const x={EncoderOnly:0,EncoderDecoder:1,Seq2Seq:2,Vision2Seq:3,DecoderOnly:4,MaskGeneration:5,ImageTextToText:6,Musicgen:7,MultiModality:8,Phi3V:9,AudioTextToText:10,AutoEncoder:11,ImageAudioTextToText:12},M=new Map,T=new Map,v=new Map;async function P(w,S,L){var Lr;let le=((Lr=L.config)==null?void 0:Lr["transformers.js_config"])??{},ve=L.device??le.device;ve&&typeof ve!="string"&&(ve.hasOwnProperty(S)?ve=ve[S]:(console.warn(`device not specified for "${S}". Using the default device.`),ve=null));const be=ve??(g.apis.IS_NODE_ENV?"cpu":"wasm"),Fe=(0,o.deviceToExecutionProviders)(be),Be=le.device_config??{};Be.hasOwnProperty(be)&&(le={...le,...Be[be]});let Ke=L.dtype??le.dtype;if(typeof Ke!="string"&&(Ke&&Ke.hasOwnProperty(S)?Ke=Ke[S]:(Ke=n.DEFAULT_DEVICE_DTYPE_MAPPING[be]??n.DATA_TYPES.fp32,console.warn(`dtype not specified for "${S}". Using the default dtype (${Ke}) for this device (${be}).`))),Ke===n.DATA_TYPES.auto){let At=le.dtype;typeof At!="string"&&(At=At==null?void 0:At[S]),At&&At!==n.DATA_TYPES.auto&&n.DATA_TYPES.hasOwnProperty(At)?Ke=At:Ke=n.DEFAULT_DEVICE_DTYPE_MAPPING[be]??n.DATA_TYPES.fp32}const Ye=Ke;if(n.DEFAULT_DTYPE_SUFFIX_MAPPING.hasOwnProperty(Ye)){if(Ye===n.DATA_TYPES.fp16&&be==="webgpu"&&!await(0,n.isWebGpuFp16Supported)())throw new Error(`The device (${be}) does not support fp16.`)}else throw new Error(`Invalid dtype: ${Ye}. Should be one of: ${Object.keys(n.DATA_TYPES).join(", ")}`);const mt=le.kv_cache_dtype,yt=mt?typeof mt=="string"?mt:mt[Ye]??"float32":void 0;if(yt&&!["float32","float16"].includes(yt))throw new Error(`Invalid kv_cache_dtype: ${yt}. Should be one of: float32, float16`);const vt={dtype:Ye,kv_cache_dtype:yt,device:be},Ot=n.DEFAULT_DTYPE_SUFFIX_MAPPING[Ye],_t=`${S}${Ot}.onnx`,$t=`${L.subfolder??""}/${_t}`,ct={...L.session_options};ct.executionProviders??(ct.executionProviders=Fe);const Ft=le.free_dimension_overrides;Ft?ct.freeDimensionOverrides??(ct.freeDimensionOverrides=Ft):be.startsWith("webnn")&&!ct.freeDimensionOverrides&&console.warn(`WebNN does not currently support dynamic shapes and requires 'free_dimension_overrides' to be set in config.json, preferably as a field within config["transformers.js_config"]["device_config"]["${be}"]. When 'free_dimension_overrides' is not set, you may experience significant performance degradation.`);const Ht=g.apis.IS_NODE_ENV&&g.env.useFSCache,tr=(0,l.getModelFile)(w,$t,!0,L,Ht),hr=L.use_external_data_format??le.use_external_data_format;let lr=[];if(hr){let At;typeof hr=="object"?hr.hasOwnProperty(_t)?At=hr[_t]:hr.hasOwnProperty(S)?At=hr[S]:At=!1:At=hr;const Pr=+At;if(Pr>l.MAX_EXTERNAL_DATA_CHUNKS)throw new Error(`The number of external data chunks (${Pr}) exceeds the maximum allowed value (${l.MAX_EXTERNAL_DATA_CHUNKS}).`);for(let as=0;as<Pr;++as){const Ln=`${_t}_data${as===0?"":"_"+as}`,Xr=`${L.subfolder??""}/${Ln}`;lr.push(new Promise(async(gs,gi)=>{const zn=await(0,l.getModelFile)(w,Xr,!0,L,Ht);gs(zn instanceof Uint8Array?{path:Ln,data:zn}:Ln)}))}}else ct.externalData!==void 0&&(lr=ct.externalData.map(async At=>{if(typeof At.data=="string"){const Pr=await(0,l.getModelFile)(w,At.data,!0,L);return{...At,data:Pr}}return At}));if(lr.length>0){const At=await Promise.all(lr);g.apis.IS_NODE_ENV||(ct.externalData=At)}if(be==="webgpu"){const At=(0,s.getCacheShapes)(L.config,{prefix:"present"});if(Object.keys(At).length>0&&!(0,o.isONNXProxy)()){const Pr={};for(const as in At)Pr[as]="gpu-buffer";ct.preferredOutputLocation=Pr}}return{buffer_or_path:await tr,session_options:ct,session_config:vt}}async function F(w,S,L){return Object.fromEntries(await Promise.all(Object.keys(S).map(async le=>{const{buffer_or_path:ve,session_options:be,session_config:Fe}=await P(w,S[le],L),Be=await(0,o.createInferenceSession)(ve,be,Fe);return[le,Be]})))}async function D(w,S,L){return Object.fromEntries(await Promise.all(Object.keys(S).map(async le=>{const ve=await(0,l.getModelJSON)(w,S[le],!1,L);return[le,ve]})))}function K(w,S){const L=Object.create(null),le=[];for(const Fe of w.inputNames){const Be=S[Fe];if(!(Be instanceof u.Tensor)){le.push(Fe);continue}L[Fe]=(0,o.isONNXProxy)()?Be.clone():Be}if(le.length>0)throw new Error(`An error occurred during model execution: "Missing the following inputs: ${le.join(", ")}.`);const ve=Object.keys(S).length,be=w.inputNames.length;if(ve>be){let Fe=Object.keys(S).filter(Be=>!w.inputNames.includes(Be));console.warn(`WARNING: Too many inputs were provided (${ve} > ${be}). The following inputs will be ignored: "${Fe.join(", ")}".`)}return L}async function U(w,S){const L=K(w,S);try{const le=Object.fromEntries(Object.entries(L).map(([be,Fe])=>[be,Fe.ort_tensor])),ve=await(0,o.runInferenceSession)(w,le);return j(ve)}catch(le){const ve=Object.fromEntries(Object.entries(L).map(([be,Fe])=>{const Be={type:Fe.type,dims:Fe.dims,location:Fe.location};return Be.location!=="gpu-buffer"&&(Be.data=Fe.data),[be,Be]}));throw console.error(`An error occurred during model execution: "${le}".`),console.error("Inputs given to model:",ve),le}}function j(w){for(let S in w)(0,o.isONNXTensor)(w[S])?w[S]=new u.Tensor(w[S]):typeof w[S]=="object"&&j(w[S]);return w}function ne(w){if(w instanceof u.Tensor)return w;if(w.length===0)throw Error("items must be non-empty");if(Array.isArray(w[0])){if(w.some(S=>S.length!==w[0].length))throw Error("Unable to create tensor, you should probably activate truncation and/or padding with 'padding=True' and/or 'truncation=True' to have batched tensors with the same length.");return new u.Tensor("int64",BigInt64Array.from(w.flat().map(S=>BigInt(S))),[w.length,w[0].length])}else return new u.Tensor("int64",BigInt64Array.from(w.map(S=>BigInt(S))),[1,w.length])}function q(w){return new u.Tensor("bool",[w],[1])}async function te(w,S){let{encoder_outputs:L,input_ids:le,decoder_input_ids:ve,...be}=S;if(!L){const Be=(0,i.pick)(S,w.sessions.model.inputNames);L=(await Z(w,Be)).last_hidden_state}return be.input_ids=ve,be.encoder_hidden_states=L,w.sessions.decoder_model_merged.inputNames.includes("encoder_attention_mask")&&(be.encoder_attention_mask=S.attention_mask),await he(w,be,!0)}async function Z(w,S){const L=w.sessions.model,le=(0,i.pick)(S,L.inputNames);if(L.inputNames.includes("inputs_embeds")&&!le.inputs_embeds){if(!S.input_ids)throw new Error("Both `input_ids` and `inputs_embeds` are missing in the model inputs.");le.inputs_embeds=await w.encode_text({input_ids:S.input_ids})}if(L.inputNames.includes("token_type_ids")&&!le.token_type_ids){if(!le.input_ids)throw new Error("Both `input_ids` and `token_type_ids` are missing in the model inputs.");le.token_type_ids=(0,u.zeros_like)(le.input_ids)}if(L.inputNames.includes("pixel_mask")&&!le.pixel_mask){if(!le.pixel_values)throw new Error("Both `pixel_values` and `pixel_mask` are missing in the model inputs.");const ve=le.pixel_values.dims;le.pixel_mask=(0,u.ones)([ve[0],ve[2],ve[3]])}return await U(L,le)}async function ae(w,S){const L=await w.encode(S);return await w.decode(L)}async function he(w,S,L=!1){const le=w.sessions[L?"decoder_model_merged":"model"],{past_key_values:ve,...be}=S;if(le.inputNames.includes("use_cache_branch")&&(be.use_cache_branch=q(!!ve)),le.inputNames.includes("position_ids")&&be.attention_mask&&!be.position_ids){const Be=["paligemma","gemma3_text","gemma3"].includes(w.config.model_type)?1:0;be.position_ids=me(be,ve,Be)}w.addPastKeyValues(be,ve);const Fe=(0,i.pick)(be,le.inputNames);return await U(le,Fe)}function Q({modality_token_id:w,inputs_embeds:S,modality_features:L,input_ids:le,attention_mask:ve}){const be=le.tolist().map(Ye=>Ye.reduce((mt,yt,vt)=>(yt==w&&mt.push(vt),mt),[])),Fe=be.reduce((Ye,mt)=>Ye+mt.length,0),Be=L.dims[0];if(Fe!==Be)throw new Error(`Number of tokens and features do not match: tokens: ${Fe}, features ${Be}`);let Ke=0;for(let Ye=0;Ye<be.length;++Ye){const mt=be[Ye],yt=S[Ye];for(let vt=0;vt<mt.length;++vt)yt[mt[vt]].data.set(L[Ke++].data)}return{inputs_embeds:S,attention_mask:ve}}function B({image_token_id:w,inputs_embeds:S,image_features:L,input_ids:le,attention_mask:ve}){return Q({modality_token_id:w,inputs_embeds:S,modality_features:L,input_ids:le,attention_mask:ve})}function O({audio_token_id:w,inputs_embeds:S,audio_features:L,input_ids:le,attention_mask:ve}){return Q({modality_token_id:w,inputs_embeds:S,modality_features:L,input_ids:le,attention_mask:ve})}async function W(w,{encode_function:S,merge_function:L,modality_input_name:le,modality_output_name:ve,input_ids:be=null,attention_mask:Fe=null,position_ids:Be=null,inputs_embeds:Ke=null,past_key_values:Ye=null,generation_config:mt=null,logits_processor:yt=null,...vt}){const Ot=vt[le];if(!Ke){if(Ke=await w.encode_text({input_ids:be,...vt}),Ot&&be.dims[1]!==1){const $t=await S({[le]:Ot,...vt});({inputs_embeds:Ke,attention_mask:Fe}=L({[ve]:$t,inputs_embeds:Ke,input_ids:be,attention_mask:Fe}))}else if(Ye&&Ot&&be.dims[1]===1){const $t=be.dims[1],ct=Object.values(Ye)[0].dims.at(-2);Fe=(0,u.cat)([(0,u.ones)([be.dims[0],ct]),Fe.slice(null,[Fe.dims[1]-$t,Fe.dims[1]])],1)}}if(!Be&&w.config.model_type==="qwen2_vl"){const{image_grid_thw:$t,video_grid_thw:ct}=vt;[Be]=w.get_rope_index(be,$t,ct,Fe)}return await he(w,{inputs_embeds:Ke,past_key_values:Ye,attention_mask:Fe,position_ids:Be,generation_config:mt,logits_processor:yt},!0)}async function N(w,S){return await W(w,{...S,modality_input_name:"audio_values",modality_output_name:"audio_features",encode_function:w.encode_audio.bind(w),merge_function:w._merge_input_ids_with_audio_features.bind(w)})}async function J(w,S){return await W(w,{...S,modality_input_name:"pixel_values",modality_output_name:"image_features",encode_function:w.encode_image.bind(w),merge_function:w._merge_input_ids_with_image_features.bind(w)})}function ie(w,S=0){const[L,le]=w.dims,ve=w.data,be=new BigInt64Array(ve.length);for(let Fe=0;Fe<L;++Fe){const Be=Fe*le;let Ke=BigInt(S);for(let Ye=0;Ye<le;++Ye){const mt=Be+Ye;ve[mt]===0n?be[mt]=BigInt(1):(be[mt]=Ke,Ke+=ve[mt])}}return{data:be,dims:w.dims}}function me(w,S=null,L=0){const{input_ids:le,inputs_embeds:ve,attention_mask:be}=w,{data:Fe,dims:Be}=ie(be,L);let Ke=new u.Tensor("int64",Fe,Be);if(S){const Ye=-(le??ve).dims.at(1);Ke=Ke.slice(null,[Ye,null])}return Ke}function Ae(w,S,L,le){const ve=L.past_key_values?Object.values(L.past_key_values)[0].dims.at(-2):0;if(!L.attention_mask){let be;for(const Fe of["input_ids","inputs_embeds","position_ids"])if(L[Fe]){be=L[Fe].dims;break}if(!be)throw new Error("attention_mask is not provided, and unable to infer its shape from model inputs.");L.attention_mask=(0,u.ones)([be[0],ve+be[1]])}if(L.past_key_values){const{input_ids:be,attention_mask:Fe}=L;Fe&&Fe.dims[1]>be.dims[1]||ve<be.dims[1]&&(L.input_ids=be.slice(null,[ve,null]))}return L}function Ve(w,S,L,le){return L.past_key_values&&(S=S.map(ve=>[ve.at(-1)])),{...L,decoder_input_ids:ne(S)}}function $e(w,...S){return w.config.is_encoder_decoder?Ve(w,...S):Ae(w,...S)}function X(w,S,L,le){const ve=!!L.past_key_values;return le.guidance_scale!==null&&le.guidance_scale>1&&(ve?L.input_ids=(0,u.cat)([L.input_ids,L.input_ids],0):(L.input_ids=(0,u.cat)([L.input_ids,(0,u.full_like)(L.input_ids,BigInt(le.pad_token_id))],0),L.attention_mask=(0,u.cat)([L.attention_mask,(0,u.full_like)(L.attention_mask,0n)],0))),(ve||!L.pixel_values)&&(L.pixel_values=(0,u.full)([0,0,3,384,384],1)),ve&&(L.images_seq_mask=new u.Tensor("bool",new Array(1).fill(!0).fill(!1,0,1),[1,1]),L.images_emb_mask=new u.Tensor("bool",new Array(0).fill(!1),[1,1,0])),L}class z extends a.Callable{constructor(L,le,ve){super();re(this,"main_input_name","input_ids");re(this,"forward_params",["input_ids","attention_mask"]);this.config=L,this.sessions=le,this.configs=ve;const be=v.get(this.constructor),Fe=M.get(be);switch(this.can_generate=!1,this._forward=null,this._prepare_inputs_for_generation=null,Fe){case x.DecoderOnly:this.can_generate=!0,this._forward=he,this._prepare_inputs_for_generation=Ae;break;case x.Seq2Seq:case x.Vision2Seq:case x.Musicgen:this.can_generate=!0,this._forward=te,this._prepare_inputs_for_generation=Ve;break;case x.EncoderDecoder:this._forward=te;break;case x.ImageTextToText:this.can_generate=!0,this._forward=J,this._prepare_inputs_for_generation=$e;break;case x.AudioTextToText:this.can_generate=!0,this._forward=N,this._prepare_inputs_for_generation=$e;break;case x.Phi3V:case x.ImageAudioTextToText:this.can_generate=!0,this._prepare_inputs_for_generation=$e;break;case x.MultiModality:this.can_generate=!0,this._prepare_inputs_for_generation=X;break;case x.AutoEncoder:this._forward=ae;break;default:this._forward=Z;break}this.can_generate&&this.forward_params.push("past_key_values"),this.custom_config=this.config["transformers.js_config"]??{}}async dispose(){var le;const L=[];for(const ve of Object.values(this.sessions))(le=ve==null?void 0:ve.handler)!=null&&le.dispose&&L.push(ve.handler.dispose());return await Promise.all(L)}static async from_pretrained(L,{progress_callback:le=null,config:ve=null,cache_dir:be=null,local_files_only:Fe=!1,revision:Be="main",model_file_name:Ke=null,subfolder:Ye="onnx",device:mt=null,dtype:yt=null,use_external_data_format:vt=null,session_options:Ot={}}={}){let _t={progress_callback:le,config:ve,cache_dir:be,local_files_only:Fe,revision:Be,model_file_name:Ke,subfolder:Ye,device:mt,dtype:yt,use_external_data_format:vt,session_options:Ot};const $t=v.get(this),ct=M.get($t);ve=_t.config=await s.AutoConfig.from_pretrained(L,_t);let Ft;if(ct===x.DecoderOnly)Ft=await Promise.all([F(L,{model:_t.model_file_name??"model"},_t),D(L,{generation_config:"generation_config.json"},_t)]);else if(ct===x.Seq2Seq||ct===x.Vision2Seq)Ft=await Promise.all([F(L,{model:"encoder_model",decoder_model_merged:"decoder_model_merged"},_t),D(L,{generation_config:"generation_config.json"},_t)]);else if(ct===x.MaskGeneration)Ft=await Promise.all([F(L,{model:"vision_encoder",prompt_encoder_mask_decoder:"prompt_encoder_mask_decoder"},_t)]);else if(ct===x.EncoderDecoder)Ft=await Promise.all([F(L,{model:"encoder_model",decoder_model_merged:"decoder_model_merged"},_t)]);else if(ct===x.ImageTextToText){const Ht={embed_tokens:"embed_tokens",vision_encoder:"vision_encoder",decoder_model_merged:"decoder_model_merged"};ve.is_encoder_decoder&&(Ht.model="encoder_model"),Ft=await Promise.all([F(L,Ht,_t),D(L,{generation_config:"generation_config.json"},_t)])}else if(ct===x.AudioTextToText){const Ht={embed_tokens:"embed_tokens",audio_encoder:"audio_encoder",decoder_model_merged:"decoder_model_merged"};Ft=await Promise.all([F(L,Ht,_t),D(L,{generation_config:"generation_config.json"},_t)])}else if(ct===x.ImageAudioTextToText){const Ht={embed_tokens:"embed_tokens",audio_encoder:"audio_encoder",vision_encoder:"vision_encoder",decoder_model_merged:"decoder_model_merged"};Ft=await Promise.all([F(L,Ht,_t),D(L,{generation_config:"generation_config.json"},_t)])}else if(ct===x.Musicgen)Ft=await Promise.all([F(L,{model:"text_encoder",decoder_model_merged:"decoder_model_merged",encodec_decode:"encodec_decode"},_t),D(L,{generation_config:"generation_config.json"},_t)]);else if(ct===x.MultiModality)Ft=await Promise.all([F(L,{prepare_inputs_embeds:"prepare_inputs_embeds",model:"language_model",lm_head:"lm_head",gen_head:"gen_head",gen_img_embeds:"gen_img_embeds",image_decode:"image_decode"},_t),D(L,{generation_config:"generation_config.json"},_t)]);else if(ct===x.Phi3V)Ft=await Promise.all([F(L,{prepare_inputs_embeds:"prepare_inputs_embeds",model:"model",vision_encoder:"vision_encoder"},_t),D(L,{generation_config:"generation_config.json"},_t)]);else if(ct===x.AutoEncoder)Ft=await Promise.all([F(L,{encoder_model:"encoder_model",decoder_model:"decoder_model"},_t)]);else{if(ct!==x.EncoderOnly){const Ht=$t??(ve==null?void 0:ve.model_type);Ht!=="custom"&&console.warn(`Model type for '${Ht}' not found, assuming encoder-only architecture. Please report this at ${c.GITHUB_ISSUE_URL}.`)}Ft=await Promise.all([F(L,{model:_t.model_file_name??"model"},_t)])}return new this(ve,...Ft)}async _call(L){return await this.forward(L)}async forward(L){return await this._forward(this,L)}get generation_config(){var L;return((L=this.configs)==null?void 0:L.generation_config)??null}_get_logits_processor(L,le,ve=null){const be=new p.LogitsProcessorList;if(L.repetition_penalty!==null&&L.repetition_penalty!==1&&be.push(new p.RepetitionPenaltyLogitsProcessor(L.repetition_penalty)),L.no_repeat_ngram_size!==null&&L.no_repeat_ngram_size>0&&be.push(new p.NoRepeatNGramLogitsProcessor(L.no_repeat_ngram_size)),L.bad_words_ids!==null&&be.push(new p.NoBadWordsLogitsProcessor(L.bad_words_ids,L.eos_token_id)),L.min_length!==null&&L.eos_token_id!==null&&L.min_length>0&&be.push(new p.MinLengthLogitsProcessor(L.min_length,L.eos_token_id)),L.min_new_tokens!==null&&L.eos_token_id!==null&&L.min_new_tokens>0&&be.push(new p.MinNewTokensLengthLogitsProcessor(le,L.min_new_tokens,L.eos_token_id)),L.forced_bos_token_id!==null&&be.push(new p.ForcedBOSTokenLogitsProcessor(L.forced_bos_token_id)),L.forced_eos_token_id!==null&&be.push(new p.ForcedEOSTokenLogitsProcessor(L.max_length,L.forced_eos_token_id)),L.begin_suppress_tokens!==null){const Fe=le>1||L.forced_bos_token_id===null?le:le+1;be.push(new p.SuppressTokensAtBeginLogitsProcessor(L.begin_suppress_tokens,Fe))}return L.guidance_scale!==null&&L.guidance_scale>1&&be.push(new p.ClassifierFreeGuidanceLogitsProcessor(L.guidance_scale)),L.temperature===0&&L.do_sample&&(console.warn("`do_sample` changed to false because `temperature: 0` implies greedy sampling (always selecting the most likely token), which is incompatible with `do_sample: true`."),L.do_sample=!1),L.do_sample&&L.temperature!==null&&L.temperature!==1&&be.push(new p.TemperatureLogitsWarper(L.temperature)),ve!==null&&be.extend(ve),be}_prepare_generation_config(L,le,ve=d.GenerationConfig){const be={...this.config};for(const Be of["decoder","generator","text_config"])Be in be&&Object.assign(be,be[Be]);const Fe=new ve(be);return Object.assign(Fe,this.generation_config??{}),L&&Object.assign(Fe,L),le&&Object.assign(Fe,(0,i.pick)(le,Object.getOwnPropertyNames(Fe))),Fe}_get_stopping_criteria(L,le=null){const ve=new b.StoppingCriteriaList;return L.max_length!==null&&ve.push(new b.MaxLengthCriteria(L.max_length,this.config.max_position_embeddings??null)),L.eos_token_id!==null&&ve.push(new b.EosTokenCriteria(L.eos_token_id)),le&&ve.extend(le),ve}_validate_model_class(){if(!this.can_generate){const L=[Wc,Gc,Uc,Vc],le=v.get(this.constructor),ve=new Set,be=this.config.model_type;for(const Be of L){const Ke=Be.get(be);Ke&&ve.add(Ke[0])}let Fe=`The current model class (${le}) is not compatible with \`.generate()\`, as it doesn't have a language model head.`;throw ve.size>0&&(Fe+=` Please use the following class instead: ${[...ve].join(", ")}`),Error(Fe)}}prepare_inputs_for_generation(...L){return this._prepare_inputs_for_generation(this,...L)}_update_model_kwargs_for_generation({generated_input_ids:L,outputs:le,model_inputs:ve,is_encoder_decoder:be}){return ve.past_key_values=this.getPastKeyValues(le,ve.past_key_values),ve.input_ids=new u.Tensor("int64",L.flat(),[L.length,1]),be||(ve.attention_mask=(0,u.cat)([ve.attention_mask,(0,u.ones)([ve.attention_mask.dims[0],1])],1)),ve.position_ids=null,ve}_prepare_model_inputs({inputs:L,bos_token_id:le,model_kwargs:ve}){const be=(0,i.pick)(ve,this.forward_params),Fe=this.main_input_name;if(Fe in be){if(L)throw new Error("`inputs`: {inputs}` were passed alongside {input_name} which is not allowed. Make sure to either pass {inputs} or {input_name}=...")}else be[Fe]=L;return{inputs_tensor:be[Fe],model_inputs:be,model_input_name:Fe}}async _prepare_encoder_decoder_kwargs_for_generation({inputs_tensor:L,model_inputs:le,model_input_name:ve,generation_config:be}){if(this.sessions.model.inputNames.includes("inputs_embeds")&&!le.inputs_embeds&&"_prepare_inputs_embeds"in this){const{input_ids:Be,pixel_values:Ke,attention_mask:Ye,...mt}=le,yt=await this._prepare_inputs_embeds(le);le={...mt,...(0,i.pick)(yt,["inputs_embeds","attention_mask"])}}let{last_hidden_state:Fe}=await Z(this,le);if(be.guidance_scale!==null&&be.guidance_scale>1)Fe=(0,u.cat)([Fe,(0,u.full_like)(Fe,0)],0),"attention_mask"in le&&(le.attention_mask=(0,u.cat)([le.attention_mask,(0,u.zeros_like)(le.attention_mask)],0));else if(le.decoder_input_ids){const Be=ne(le.decoder_input_ids).dims[0];if(Be!==Fe.dims[0]){if(Fe.dims[0]!==1)throw new Error(`The encoder outputs have a different batch size (${Fe.dims[0]}) than the decoder inputs (${Be}).`);Fe=(0,u.cat)(Array.from({length:Be},()=>Fe),0)}}return le.encoder_outputs=Fe,le}_prepare_decoder_input_ids_for_generation({batch_size:L,model_input_name:le,model_kwargs:ve,decoder_start_token_id:be,bos_token_id:Fe,generation_config:Be}){let{decoder_input_ids:Ke,...Ye}=ve;if(!(Ke instanceof u.Tensor)){if(Ke)Array.isArray(Ke[0])||(Ke=Array.from({length:L},()=>Ke));else if(be??(be=Fe),this.config.model_type==="musicgen")Ke=Array.from({length:L*this.config.decoder.num_codebooks},()=>[be]);else if(Array.isArray(be)){if(be.length!==L)throw new Error(`\`decoder_start_token_id\` expcted to have length ${L} but got ${be.length}`);Ke=be}else Ke=Array.from({length:L},()=>[be]);Ke=ne(Ke)}return ve.decoder_attention_mask=(0,u.ones_like)(Ke),{input_ids:Ke,model_inputs:Ye}}async generate({inputs:L=null,generation_config:le=null,logits_processor:ve=null,stopping_criteria:be=null,streamer:Fe=null,...Be}){this._validate_model_class(),le=this._prepare_generation_config(le,Be);let{inputs_tensor:Ke,model_inputs:Ye,model_input_name:mt}=this._prepare_model_inputs({inputs:L,model_kwargs:Be});const yt=this.config.is_encoder_decoder;yt&&("encoder_outputs"in Ye||(Ye=await this._prepare_encoder_decoder_kwargs_for_generation({inputs_tensor:Ke,model_inputs:Ye,model_input_name:mt,generation_config:le})));let vt;yt?{input_ids:vt,model_inputs:Ye}=this._prepare_decoder_input_ids_for_generation({batch_size:Ye[mt].dims.at(0),model_input_name:mt,model_kwargs:Ye,decoder_start_token_id:le.decoder_start_token_id,bos_token_id:le.bos_token_id,generation_config:le}):vt=Ye[mt];let Ot=vt.dims.at(-1);le.max_new_tokens!==null&&(le.max_length=Ot+le.max_new_tokens);const _t=this._get_logits_processor(le,Ot,ve),$t=this._get_stopping_criteria(le,be),ct=Ye[mt].dims.at(0),Ft=A.LogitsSampler.getSampler(le),Ht=new Array(ct).fill(0),tr=vt.tolist();Fe&&Fe.put(tr);let hr,lr={};for(;;){if(Ye=this.prepare_inputs_for_generation(tr,Ye,le),hr=await this.forward(Ye),le.output_attentions&&le.return_dict_in_generate){const Xr=this.getAttentions(hr);for(const gs in Xr)gs in lr||(lr[gs]=[]),lr[gs].push(Xr[gs])}const At=hr.logits.slice(null,-1,null),Pr=_t(tr,At),as=[];for(let Xr=0;Xr<Pr.dims.at(0);++Xr){const gs=Pr[Xr],gi=await Ft(gs);for(const[zn,wi]of gi){const Mi=BigInt(zn);Ht[Xr]+=wi,tr[Xr].push(Mi),as.push([Mi]);break}}if(Fe&&Fe.put(as),$t(tr).every(Xr=>Xr))break;Ye=this._update_model_kwargs_for_generation({generated_input_ids:as,outputs:hr,model_inputs:Ye,is_encoder_decoder:yt})}Fe&&Fe.end();const br=this.getPastKeyValues(hr,Ye.past_key_values,!0),Lr=new u.Tensor("int64",tr.flat(),[tr.length,tr[0].length]);if(le.return_dict_in_generate)return{sequences:Lr,past_key_values:br,...lr};for(const At of Object.values(hr))At.location==="gpu-buffer"&&At.dispose();return Lr}getPastKeyValues(L,le,ve=!1){const be=Object.create(null);for(const Fe in L)if(Fe.startsWith("present")){const Be=Fe.replace("present_conv","past_conv").replace("present","past_key_values"),Ke=Fe.includes("encoder");if(Ke&&le?be[Be]=le[Be]:be[Be]=L[Fe],le&&(!Ke||ve)){const Ye=le[Be];Ye.location==="gpu-buffer"&&Ye.dispose()}}return be}getAttentions(L){const le={};for(const ve of["cross_attentions","encoder_attentions","decoder_attentions"])for(const be in L)be.startsWith(ve)&&(ve in le||(le[ve]=[]),le[ve].push(L[be]));return le}addPastKeyValues(L,le){var ve,be,Fe;if(le)Object.assign(L,le);else{const Be=this.sessions.decoder_model_merged??this.sessions.model,Ke=((be=(ve=L[this.main_input_name]??L.attention_mask)==null?void 0:ve.dims)==null?void 0:be[0])??1,Ye=((Fe=Be==null?void 0:Be.config)==null?void 0:Fe.kv_cache_dtype)??"float32",mt=Ye==="float16"?u.DataTypeMap.float16:u.DataTypeMap.float32,yt=(0,s.getCacheShapes)(this.config,{batch_size:Ke});for(const vt in yt){const Ot=yt[vt].reduce((_t,$t)=>_t*$t,1);L[vt]=new u.Tensor(Ye,new mt(Ot),yt[vt])}}}async encode_image({pixel_values:L}){return(await U(this.sessions.vision_encoder,{pixel_values:L})).image_features}async encode_text({input_ids:L}){return(await U(this.sessions.embed_tokens,{input_ids:L})).inputs_embeds}async encode_audio({audio_values:L}){return(await U(this.sessions.audio_encoder,{audio_values:L})).audio_features}}class _e{}class Ee extends _e{constructor({last_hidden_state:S,hidden_states:L=null,attentions:le=null}){super(),this.last_hidden_state=S,this.hidden_states=L,this.attentions=le}}class Me extends z{}class Ce extends Me{}class ye extends Me{async _call(S){return new Dr(await super._call(S))}}class de extends Me{async _call(S){return new Et(await super._call(S))}}class we extends Me{async _call(S){return new Er(await super._call(S))}}class ce extends Me{async _call(S){return new Nr(await super._call(S))}}class ke extends z{}class Le extends ke{}class Te extends ke{async _call(S){return new Dr(await super._call(S))}}class We extends ke{async _call(S){return new Et(await super._call(S))}}class qe extends ke{async _call(S){return new Er(await super._call(S))}}class st extends ke{async _call(S){return new Nr(await super._call(S))}}class Ze extends z{}class ze extends Ze{}class He extends Ze{async _call(S){return new Dr(await super._call(S))}}class gt extends Ze{async _call(S){return new Et(await super._call(S))}}class dt extends Ze{async _call(S){return new Er(await super._call(S))}}class kt extends z{}class ht extends kt{}class yr extends kt{}class $r extends z{}class Vr extends $r{}class Ur extends z{}class sr extends Ur{}class Ar extends Ur{async _call(S){return new Dr(await super._call(S))}}class rn extends Ur{async _call(S){return new Et(await super._call(S))}}class sn extends Ur{async _call(S){return new Er(await super._call(S))}}class nn extends Ur{async _call(S){return new Nr(await super._call(S))}}class ds extends z{}class ft extends ds{}class Os extends ds{async _call(S){return new Dr(await super._call(S))}}class Ds extends ds{async _call(S){return new Et(await super._call(S))}}class St extends ds{async _call(S){return new Er(await super._call(S))}}class Kt extends ds{async _call(S){return new Nr(await super._call(S))}}class $ extends z{}class ee extends ${}class V extends ${async _call(S){return new Dr(await super._call(S))}}class Y extends ${async _call(S){return new Et(await super._call(S))}}class oe extends ${async _call(S){return new Er(await super._call(S))}}class xe extends ${async _call(S){return new Nr(await super._call(S))}}class De extends z{}class nt extends De{}class wt extends De{async _call(S){return new Dr(await super._call(S))}}class pt extends De{async _call(S){return new Et(await super._call(S))}}class xt extends De{async _call(S){return new Er(await super._call(S))}}class tt extends De{async _call(S){return new Nr(await super._call(S))}}class It extends z{}class qt extends It{}class Wr extends It{async _call(S){return new Dr(await super._call(S))}}class qr extends It{async _call(S){return new Et(await super._call(S))}}class nr extends It{async _call(S){return new Er(await super._call(S))}}class kr extends It{async _call(S){return new Nr(await super._call(S))}}class cr extends z{}class ps extends cr{}class hs extends cr{async _call(S){return new Dr(await super._call(S))}}class Ir extends cr{async _call(S){return new Et(await super._call(S))}}class Ls extends cr{async _call(S){return new Er(await super._call(S))}}class zs extends cr{async _call(S){return new Nr(await super._call(S))}}class vr extends z{}class ms extends vr{}class Yr extends vr{async _call(S){return new Et(await super._call(S))}}class Rr extends vr{async _call(S){return new Er(await super._call(S))}}class Cs extends vr{async _call(S){return new Nr(await super._call(S))}}class mr extends vr{async _call(S){return new Dr(await super._call(S))}}class ar extends z{}class fs extends ar{}class Gs extends ar{async _call(S){return new Dr(await super._call(S))}}class Gr extends ar{async _call(S){return new Et(await super._call(S))}}class Ne extends ar{async _call(S){return new Er(await super._call(S))}}class je extends z{}class rt extends je{}class Qt extends je{async _call(S){return new Dr(await super._call(S))}}class Hs extends je{async _call(S){return new Et(await super._call(S))}}class Ss extends je{async _call(S){return new Nr(await super._call(S))}}class ss extends z{}class Tn extends ss{}class En extends ss{async _call(S){return new Dr(await super._call(S))}}class Pn extends ss{async _call(S){return new Et(await super._call(S))}}class Cn extends ss{async _call(S){return new Er(await super._call(S))}}class ge extends ss{async _call(S){return new Nr(await super._call(S))}}class k extends z{}class G extends k{}class se extends k{async _call(S){return new Dr(await super._call(S))}}class ue extends k{async _call(S){return new Et(await super._call(S))}}class fe extends k{async _call(S){return new Nr(await super._call(S))}}class Se extends z{}class Ge extends Se{}class Xe extends Se{async _call(S){return new Et(await super._call(S))}}class Qe extends Se{async _call(S){return new Nr(await super._call(S))}}class et extends Se{async _call(S){return new Dr(await super._call(S))}}class bt extends z{constructor(){super(...arguments);re(this,"forward_params",["input_ids","attention_mask","encoder_outputs","decoder_input_ids","decoder_attention_mask","past_key_values"])}}class Rt extends bt{}class Lt extends bt{}class Zt extends z{}class Gt extends Zt{}class _r extends Zt{}class gr extends z{}class dr extends gr{}class wr extends gr{}class xr extends z{}class ns extends xr{}class Yt extends xr{}class Mr extends xr{async _call(S){return new Et(await super._call(S))}}class Fr extends z{}class Zr extends Fr{}class $s extends Fr{}class Or extends Fr{async _call(S){return new Et(await super._call(S))}}class _s extends Fr{}class ir extends z{}class pr extends ir{}class fr extends ir{}class er extends z{}class Qr extends er{}class Ks extends er{}class Rs extends z{}class Ca extends Rs{}class Sa extends Rs{async _call(S){return new Dr(await super._call(S))}}class Ao extends Rs{async _call(S){return new Et(await super._call(S))}}class $a extends Rs{async _call(S){return new Er(await super._call(S))}}class Aa extends Rs{async _call(S){return new Nr(await super._call(S))}}class on extends z{}class ka extends on{}class Ia extends on{async _call(S){return new Dr(await super._call(S))}}class Fa extends on{async _call(S){return new Et(await super._call(S))}}class Oa extends on{async _call(S){return new Er(await super._call(S))}}class Da extends on{async _call(S){return new Nr(await super._call(S))}}class an extends z{}class La extends an{}class za extends an{async _call(S){return new Dr(await super._call(S))}}class Ra extends an{async _call(S){return new Et(await super._call(S))}}class ko extends an{async _call(S){return new Er(await super._call(S))}}class Io extends an{async _call(S){return new Nr(await super._call(S))}}class Qn extends z{}class Ba extends Qn{}class Na extends Qn{}class Xn extends z{constructor(){super(...arguments);re(this,"requires_attention_mask",!1);re(this,"main_input_name","input_features");re(this,"forward_params",["input_features","attention_mask","decoder_input_ids","decoder_attention_mask","past_key_values"])}}class Fo extends Xn{}class qs extends Xn{_prepare_generation_config(S,L){return super._prepare_generation_config(S,L,y.WhisperGenerationConfig)}_retrieve_init_tokens(S){const L=[S.decoder_start_token_id];let le=S.language;const ve=S.task;if(S.is_multilingual){le||(console.warn("No language specified - defaulting to English (en)."),le="en");const Fe=`<|${(0,C.whisper_language_to_code)(le)}|>`;L.push(S.lang_to_id[Fe]),L.push(S.task_to_id[ve??"transcribe"])}else if(le||ve)throw new Error("Cannot specify `task` or `language` for an English-only model. If the model is intended to be multilingual, pass `is_multilingual=true` to generate, or update the generation config.");return!S.return_timestamps&&S.no_timestamps_token_id&&L.at(-1)!==S.no_timestamps_token_id?L.push(S.no_timestamps_token_id):S.return_timestamps&&L.at(-1)===S.no_timestamps_token_id&&(console.warn("<|notimestamps|> prompt token is removed from generation_config since `return_timestamps` is set to `true`."),L.pop()),L.filter(be=>be!=null)}async generate({inputs:S=null,generation_config:L=null,logits_processor:le=null,stopping_criteria:ve=null,...be}){L=this._prepare_generation_config(L,be);const Fe=be.decoder_input_ids??this._retrieve_init_tokens(L);if(L.return_timestamps&&(le??(le=new p.LogitsProcessorList),le.push(new p.WhisperTimeStampLogitsProcessor(L,Fe))),L.begin_suppress_tokens&&(le??(le=new p.LogitsProcessorList),le.push(new p.SuppressTokensAtBeginLogitsProcessor(L.begin_suppress_tokens,Fe.length))),L.return_token_timestamps){if(!L.alignment_heads)throw new Error("Model generation config has no `alignment_heads`, token-level timestamps not available. See https://gist.github.com/hollance/42e32852f24243b748ae6bc1f985b13a on how to add this property to the generation config.");L.task==="translate"&&console.warn("Token-level timestamps may not be reliable for task 'translate'."),L.output_attentions=!0,L.return_dict_in_generate=!0}const Be=await super.generate({inputs:S,generation_config:L,logits_processor:le,decoder_input_ids:Fe,...be});return L.return_token_timestamps&&(Be.token_timestamps=this._extract_token_timestamps(Be,L.alignment_heads,L.num_frames)),Be}_extract_token_timestamps(S,L,le=null,ve=.02){if(!S.cross_attentions)throw new Error("Model outputs must contain cross attentions to extract timestamps. This is most likely because the model was not exported with `output_attentions=True`.");le==null&&console.warn("`num_frames` has not been set, meaning the entire audio will be analyzed. This may lead to inaccurate token-level timestamps for short audios (< 30 seconds).");let be=this.config.median_filter_width;be===void 0&&(console.warn("Model config has no `median_filter_width`, using default value of 7."),be=7);const Fe=S.cross_attentions,Be=Array.from({length:this.config.decoder_layers},($t,ct)=>(0,u.cat)(Fe.map(Ft=>Ft[ct]),2)),Ke=(0,u.stack)(L.map(([$t,ct])=>{if($t>=Be.length)throw new Error(`Layer index ${$t} is out of bounds for cross attentions (length ${Be.length}).`);return le?Be[$t].slice(null,ct,null,[0,le]):Be[$t].slice(null,ct)})).transpose(1,0,2,3),[Ye,mt]=(0,u.std_mean)(Ke,-2,0,!0),yt=Ke.clone();for(let $t=0;$t<yt.dims[0];++$t){const ct=yt[$t];for(let Ft=0;Ft<ct.dims[0];++Ft){const Ht=ct[Ft],tr=Ye[$t][Ft][0].data,hr=mt[$t][Ft][0].data;for(let lr=0;lr<Ht.dims[0];++lr){let br=Ht[lr].data;for(let Lr=0;Lr<br.length;++Lr)br[Lr]=(br[Lr]-hr[Lr])/tr[Lr];br.set((0,f.medianFilter)(br,be))}}}const vt=[(0,u.mean)(yt,1)],Ot=S.sequences.dims,_t=new u.Tensor("float32",new Float32Array(Ot[0]*Ot[1]),Ot);for(let $t=0;$t<Ot[0];++$t){const ct=vt[$t].neg().squeeze_(0),[Ft,Ht]=(0,f.dynamic_time_warping)(ct.tolist()),tr=Array.from({length:Ft.length-1},(br,Lr)=>Ft[Lr+1]-Ft[Lr]),hr=(0,i.mergeArrays)([1],tr).map(br=>!!br),lr=[];for(let br=0;br<hr.length;++br)hr[br]&&lr.push(Ht[br]*ve);_t[$t].data.set(lr,1)}return _t}}class ja extends qs{}class Jn extends z{constructor(){super(...arguments);re(this,"requires_attention_mask",!1);re(this,"main_input_name","input_values");re(this,"forward_params",["input_values","decoder_input_ids","past_key_values"])}}class $u extends Jn{}class Va extends Jn{}class Oo extends z{constructor(){super(...arguments);re(this,"main_input_name","pixel_values");re(this,"forward_params",["pixel_values","decoder_input_ids","encoder_hidden_states","past_key_values"])}}class Sn extends z{constructor(){super(...arguments);re(this,"forward_params",["input_ids","attention_mask","pixel_values","position_ids","past_key_values"])}}class $n extends Sn{_merge_input_ids_with_image_features(S){const L=S.image_features.dims.at(-1),le=S.image_features.view(-1,L);return B({image_token_id:this.config.image_token_index,...S,image_features:le})}}class Ua extends $n{}class Wa extends $n{}class Ga extends z{constructor(){super(...arguments);re(this,"forward_params",["input_ids","inputs_embeds","attention_mask","pixel_values","encoder_outputs","decoder_input_ids","decoder_inputs_embeds","decoder_attention_mask","past_key_values"]);re(this,"main_input_name","inputs_embeds")}}class Ha extends Ga{_merge_input_ids_with_image_features({inputs_embeds:S,image_features:L,input_ids:le,attention_mask:ve}){return{inputs_embeds:(0,u.cat)([L,S],1),attention_mask:(0,u.cat)([(0,u.ones)(L.dims.slice(0,2)),ve],1)}}async _prepare_inputs_embeds({input_ids:S,pixel_values:L,inputs_embeds:le,attention_mask:ve}){if(!S&&!L)throw new Error("Either `input_ids` or `pixel_values` should be provided.");let be,Fe;return S&&(be=await this.encode_text({input_ids:S})),L&&(Fe=await this.encode_image({pixel_values:L})),be&&Fe?{inputs_embeds:le,attention_mask:ve}=this._merge_input_ids_with_image_features({inputs_embeds:be,image_features:Fe,input_ids:S,attention_mask:ve}):le=be||Fe,{inputs_embeds:le,attention_mask:ve}}async forward({input_ids:S,pixel_values:L,attention_mask:le,decoder_input_ids:ve,decoder_attention_mask:be,encoder_outputs:Fe,past_key_values:Be,inputs_embeds:Ke,decoder_inputs_embeds:Ye}){if(Ke||({inputs_embeds:Ke,attention_mask:le}=await this._prepare_inputs_embeds({input_ids:S,pixel_values:L,inputs_embeds:Ke,attention_mask:le})),!Fe){let{last_hidden_state:vt}=await Z(this,{inputs_embeds:Ke,attention_mask:le});Fe=vt}if(!Ye){if(!ve)throw new Error("Either `decoder_input_ids` or `decoder_inputs_embeds` should be provided.");Ye=await this.encode_text({input_ids:ve})}return await he(this,{inputs_embeds:Ye,attention_mask:be,encoder_attention_mask:le,encoder_hidden_states:Fe,past_key_values:Be},!0)}}class Ka extends z{constructor(){super(...arguments);re(this,"forward_params",["input_ids","attention_mask","pixel_values","position_ids","past_key_values"])}}class qa extends Ka{_merge_input_ids_with_image_features(S){const L=S.image_features.dims.at(-1),le=S.image_features.view(-1,L);return B({image_token_id:this.config.image_token_index,...S,image_features:le})}}class Qa extends Sn{_merge_input_ids_with_image_features(S){const L=S.image_features.dims.at(-1),le=S.image_features.view(-1,L);return B({image_token_id:this.config.image_token_index,...S,image_features:le})}}class Xa extends z{constructor(){super(...arguments);re(this,"forward_params",["input_ids","attention_mask","inputs_embeds","per_layer_inputs","position_ids","pixel_values","input_features","input_features_mask","past_key_values"])}}class Do extends Xa{async forward({input_ids:S=null,attention_mask:L=null,pixel_values:le=null,input_features:ve=null,input_features_mask:be=null,position_ids:Fe=null,inputs_embeds:Be=null,per_layer_inputs:Ke=null,past_key_values:Ye=null,generation_config:mt=null,logits_processor:yt=null,...vt}){if((!Be||!Ke)&&({inputs_embeds:Be,per_layer_inputs:Ke}=await U(this.sessions.embed_tokens,{input_ids:S}),S.dims[1]!==1)){if(le){const{image_features:_t}=await U(this.sessions.vision_encoder,{pixel_values:le});({inputs_embeds:Be,attention_mask:L}=this._merge_input_ids_with_image_features({image_features:_t,inputs_embeds:Be,input_ids:S,attention_mask:L}))}if(ve){const{audio_features:_t}=await U(this.sessions.audio_encoder,{input_features:ve,input_features_mask:be});({inputs_embeds:Be,attention_mask:L}=this._merge_input_ids_with_audio_features({audio_features:_t,inputs_embeds:Be,input_ids:S,attention_mask:L}))}}return await he(this,{inputs_embeds:Be,per_layer_inputs:Ke,past_key_values:Ye,attention_mask:L,position_ids:Fe,generation_config:mt,logits_processor:yt},!0)}_merge_input_ids_with_image_features(S){const L=S.image_features.dims.at(-1),le=S.image_features.view(-1,L);return B({image_token_id:this.config.image_token_id,...S,image_features:le})}_merge_input_ids_with_audio_features(S){const L=S.audio_features.dims.at(-1),le=S.audio_features.view(-1,L);return O({audio_token_id:this.config.audio_token_id,...S,audio_features:le})}}class An extends z{constructor(){super(...arguments);re(this,"forward_params",["input_ids","attention_mask","pixel_values","pixel_attention_mask","position_ids","past_key_values"])}}class Bs extends An{async encode_image({pixel_values:S,pixel_attention_mask:L}){return(await U(this.sessions.vision_encoder,{pixel_values:S,pixel_attention_mask:L})).image_features}_merge_input_ids_with_image_features(S){const L=S.image_features.dims.at(-1),le=S.image_features.view(-1,L);return B({image_token_id:this.config.image_token_id,...S,image_features:le})}}class Yn extends Bs{}class Ja extends z{constructor(){super(...arguments);re(this,"forward_params",["input_ids","inputs_embeds","attention_mask","position_ids","pixel_values","image_sizes","past_key_values"])}}class Lo extends Ja{async forward({input_ids:S=null,attention_mask:L=null,pixel_values:le=null,image_sizes:ve=null,position_ids:be=null,inputs_embeds:Fe=null,past_key_values:Be=null,generation_config:Ke=null,logits_processor:Ye=null,...mt}){if(!Fe){let vt;if(le&&S.dims[1]!==1){if(!ve)throw new Error("`image_sizes` must be provided when `pixel_values` is provided.");({image_features:vt}=await U(this.sessions.vision_encoder,{pixel_values:le,image_sizes:ve}))}else{const Ot=this.config.normalized_config.hidden_size;vt=new u.Tensor("float32",[],[0,Ot])}({inputs_embeds:Fe}=await U(this.sessions.prepare_inputs_embeds,{input_ids:S,image_features:vt}))}return await he(this,{inputs_embeds:Fe,past_key_values:Be,attention_mask:L,position_ids:be,generation_config:Ke,logits_processor:Ye},!1)}}class As extends z{}class Zn extends As{}class Ya extends As{static async from_pretrained(S,L={}){return super.from_pretrained(S,{...L,model_file_name:L.model_file_name??"text_model"})}}class zo extends As{static async from_pretrained(S,L={}){return super.from_pretrained(S,{...L,model_file_name:L.model_file_name??"text_model"})}}class Za extends As{static async from_pretrained(S,L={}){return super.from_pretrained(S,{...L,model_file_name:L.model_file_name??"vision_model"})}}class Ro extends As{static async from_pretrained(S,L={}){return super.from_pretrained(S,{...L,model_file_name:L.model_file_name??"vision_model"})}}class eo extends z{}class Bo extends eo{}class ei extends eo{static async from_pretrained(S,L={}){return super.from_pretrained(S,{...L,model_file_name:L.model_file_name??"text_model"})}}class No extends As{static async from_pretrained(S,L={}){return super.from_pretrained(S,{...L,model_file_name:L.model_file_name??"vision_model"})}}class jo extends z{}class kn extends jo{}class ut extends z{}class ti extends ut{async forward(S){const L=!S.input_ids,le=!S.pixel_values;if(L&&le)throw new Error("Either `input_ids` or `pixel_values` should be provided.");if(L&&(S.input_ids=(0,u.ones)([S.pixel_values.dims[0],1])),le){const{image_size:Ye}=this.config.vision_config;S.pixel_values=(0,u.full)([0,3,Ye,Ye],0)}const{text_embeddings:ve,image_embeddings:be,l2norm_text_embeddings:Fe,l2norm_image_embeddings:Be}=await super.forward(S),Ke={};return L||(Ke.text_embeddings=ve,Ke.l2norm_text_embeddings=Fe),le||(Ke.image_embeddings=be,Ke.l2norm_image_embeddings=Be),Ke}}class Vo extends ut{static async from_pretrained(S,L={}){return super.from_pretrained(S,{...L,model_file_name:L.model_file_name??"text_model"})}}class Uo extends ut{static async from_pretrained(S,L={}){return super.from_pretrained(S,{...L,model_file_name:L.model_file_name??"vision_model"})}}class ln extends z{}class os extends ln{}class In extends ln{}class Fn extends z{}class Wo extends Fn{}class Go extends Fn{}class to extends z{}class ro extends to{}class Ho extends to{}class so extends z{}class Ko extends so{}class On extends so{}class Dn extends z{}class no extends Dn{}class qo extends Dn{}class oo extends z{}class Qo extends oo{}class Xo extends oo{}class ao extends z{}class Jo extends ao{}class h extends ao{}class E extends z{}class I extends E{}class R extends E{}class H extends z{}class pe extends H{}class Ie extends H{}class Re extends z{}class Je extends Re{}class ot extends z{}class Tt extends ot{}class zt extends ot{}class Ut extends z{}class Tr extends Ut{}class Qs extends Ut{}class Yo extends z{}class Iw extends Yo{}class Fw extends Yo{}class Au extends z{}class Ow extends Au{}class Dw extends Au{}class ku extends z{}class Lw extends ku{}class zw extends ku{}class Iu extends z{}class Rw extends Iu{}class Bw extends Iu{}class Fu extends z{}class Nw extends Fu{}class jw extends Fu{}class Ou extends z{}class Vw extends Ou{}class Uw extends Ou{}class Du extends z{}class Ww extends Du{}class Gw extends Du{}class Lu extends z{}class Hw extends Lu{}class Kw extends Lu{}class zu extends z{}class qw extends zu{}class Qw extends zu{}class Ru extends z{}class Xw extends Ru{}class Jw extends Ru{}class Bu extends z{}class Yw extends Bu{}class Zw extends Bu{}class Nu extends z{}class eM extends Nu{}class tM extends Nu{}class ju extends z{}class rM extends ju{}class sM extends ju{}class Vu extends z{}class nM extends Vu{}class oM extends Vu{}class Uu extends z{}class aM extends Uu{}class iM extends Uu{}class Wu extends z{}class lM extends Wu{}class uM extends Wu{}class Gu extends z{}class cM extends Gu{}class dM extends Gu{}class Hu extends z{}class pM extends Hu{}class hM extends Hu{}class mM extends z{constructor(){super(...arguments);re(this,"forward_params",["input_ids","attention_mask","position_ids","past_key_values","pixel_values","image_grid_thw"])}}class fM extends mM{get_rope_index(S,L,le,ve){const{vision_config:be,image_token_id:Fe,video_token_id:Be,vision_start_token_id:Ke}=this.config,Ye=be.spatial_merge_size??2,mt=[];if(L||le){let yt=S.tolist();ve||(ve=(0,u.ones_like)(S));const vt=ve.tolist(),Ot=Array.from({length:3},Ht=>Array.from({length:S.dims[0]},tr=>Array.from({length:S.dims[1]},hr=>1))),_t=L?L.tolist():[],$t=le?le.tolist():[];let ct=0,Ft=0;for(let Ht=0;Ht<yt.length;++Ht){const tr=yt[Ht].filter((rr,Cr)=>vt[Ht][Cr]==1),lr=tr.reduce((rr,Cr,dn)=>(Cr==Ke&&rr.push(dn),rr),[]).map(rr=>tr[rr+1]),br=lr.filter(rr=>rr==Fe).length,Lr=lr.filter(rr=>rr==Be).length;let At=[],Pr=0,as=br,Ln=Lr;for(let rr=0;rr<lr.length;++rr){const Cr=tr.findIndex((Bn,ks)=>ks>Pr&&Bn==Fe),dn=tr.findIndex((Bn,ks)=>ks>Pr&&Bn==Be),Rn=as>0&&Cr!==-1?Cr:tr.length+1,uo=Ln>0&&dn!==-1?dn:tr.length+1;let bi,Kc,qc,Qc;Rn<uo?([Kc,qc,Qc]=_t[ct],++ct,--as,bi=Rn):([Kc,qc,Qc]=$t[Ft],++Ft,--Ln,bi=uo);const[FE,Xc,yi]=[Number(Kc),Math.floor(Number(qc)/Ye),Math.floor(Number(Qc)/Ye)],Jc=bi-Pr,nx=At.length>0?(0,f.max)(At.at(-1))[0]+1:0;At.push(Array.from({length:3*Jc},(Bn,ks)=>nx+ks%Jc));const Yc=Jc+nx,vi=FE*Xc*yi,OE=Array.from({length:vi},(Bn,ks)=>Yc+Math.floor(ks/(Xc*yi))),DE=Array.from({length:vi},(Bn,ks)=>Yc+Math.floor(ks/yi)%Xc),LE=Array.from({length:vi},(Bn,ks)=>Yc+ks%yi);At.push([OE,DE,LE].flat()),Pr=bi+vi}if(Pr<tr.length){const rr=At.length>0?(0,f.max)(At.at(-1))[0]+1:0,Cr=tr.length-Pr;At.push(Array.from({length:3*Cr},(dn,Rn)=>rr+Rn%Cr))}const Xr=At.reduce((rr,Cr)=>rr+Cr.length,0),gs=new Array(Xr);let gi=0;for(let rr=0;rr<3;++rr)for(let Cr=0;Cr<At.length;++Cr){const dn=At[Cr],Rn=dn.length/3;for(let uo=rr*Rn;uo<(rr+1)*Rn;++uo)gs[gi++]=dn[uo]}let zn=0;const wi=vt[Ht];for(let rr=0;rr<wi.length;++rr)if(wi[rr]==1){for(let Cr=0;Cr<3;++Cr)Ot[Cr][Ht][rr]=gs[Cr*Xr/3+zn];++zn}const Mi=(0,f.max)(gs)[0];mt.push(Mi+1-yt[Ht].length)}return[new u.Tensor("int64",Ot.flat(1/0),[3,S.dims[0],S.dims[1]]),new u.Tensor("int64",mt,[mt.length,1])]}else if(ve){const{data:yt,dims:vt}=ie(ve),Ot=BigInt64Array.from({length:3*yt.length},($t,ct)=>yt[ct%yt.length]),_t=Array.from({length:vt[0]},($t,ct)=>(0,f.max)(yt.subarray(vt[1]*ct,vt[1]*(ct+1)))[0]+1n+BigInt(vt[1]));return[new u.Tensor("int64",Ot,[3,...vt]),new u.Tensor("int64",_t,[_t.length,1])]}else{const[yt,vt]=S.dims,Ot=BigInt64Array.from({length:3*yt*vt},(_t,$t)=>BigInt(Math.floor($t%vt/yt)));return[new u.Tensor("int64",Ot,[3,...S.dims]),(0,u.zeros)([yt,1])]}}async encode_image({pixel_values:S,image_grid_thw:L}){return(await U(this.sessions.vision_encoder,{pixel_values:S,grid_thw:L})).image_features}_merge_input_ids_with_image_features(S){return B({image_token_id:this.config.image_token_id,...S})}prepare_inputs_for_generation(S,L,le){if(L.attention_mask&&!L.position_ids)if(!L.past_key_values)[L.position_ids,L.rope_deltas]=this.get_rope_index(L.input_ids,L.image_grid_thw,L.video_grid_thw,L.attention_mask);else{L.pixel_values=null;const ve=BigInt(Object.values(L.past_key_values)[0].dims.at(-2)),be=L.rope_deltas.map(Fe=>ve+Fe);L.position_ids=(0,u.stack)([be,be,be],0)}return L}}class Ku extends z{}class _M extends Ku{}class gM extends Ku{}class qu extends z{}class wM extends qu{}class MM extends qu{}class Qu extends z{}class bM extends Qu{}class yM extends Qu{}class Xu extends z{}class vM extends Xu{}class xM extends Xu{}class Ju extends z{}class TM extends Ju{}class EM extends Ju{}class Yu extends z{}class PM extends Yu{}class CM extends Yu{async _call(S){return new Et(await super._call(S))}}class Zu extends z{}class SM extends Zu{}class $M extends Zu{async _call(S){return new Et(await super._call(S))}}class AM extends z{}class kM extends AM{}class ec extends z{}class IM extends ec{}class FM extends ec{async _call(S){return new Et(await super._call(S))}}class OM extends z{}class DM extends OM{}class tc extends z{}class LM extends tc{}class zM extends tc{async _call(S){return new Et(await super._call(S))}}class RM extends z{}class BM extends RM{}class rc extends z{}class NM extends rc{}class jM extends rc{async _call(S){return new Et(await super._call(S))}}class VM extends z{}class UM extends VM{async _call(S){return new rx(await super._call(S))}}class sc extends z{}class WM extends sc{}class GM extends sc{async _call(S){return new Et(await super._call(S))}}class nc extends z{}class HM extends nc{}class KM extends nc{async _call(S){return new Et(await super._call(S))}}class oc extends z{}class qM extends oc{}class QM extends oc{}class ac extends z{}class XM extends ac{}class JM extends ac{}class ic extends z{}class YM extends ic{}class ZM extends ic{async _call(S){return new Et(await super._call(S))}}class ri extends z{}class eb extends ri{}class tb extends ri{async _call(S){return new uc(await super._call(S))}}class lc extends ri{async _call(S){return new rb(await super._call(S))}}class uc extends _e{constructor({logits:S,pred_boxes:L}){super(),this.logits=S,this.pred_boxes=L}}class rb extends _e{constructor({logits:S,pred_boxes:L,pred_masks:le}){super(),this.logits=S,this.pred_boxes=L,this.pred_masks=le}}class cc extends z{}class sb extends cc{}class nb extends cc{async _call(S){return new Zo(await super._call(S))}}class Zo extends _e{constructor({logits:S,pred_boxes:L}){super(),this.logits=S,this.pred_boxes=L}}class dc extends z{}class ob extends dc{}class ab extends dc{async _call(S){return new ib(await super._call(S))}}class ib extends Zo{}class pc extends z{}class lb extends pc{}class ub extends pc{async _call(S){return new cb(await super._call(S))}}class cb extends Zo{}class hc extends z{}class db extends hc{}class pb extends hc{async _call(S){return new Zo(await super._call(S))}}class mc extends z{}class hb extends mc{}class mb extends mc{async _call(S){return new fb(await super._call(S))}}class fb extends uc{}class fc extends z{}class _b extends fc{}class gb extends fc{async _call(S){return new Et(await super._call(S))}}class _c extends z{}class wb extends _c{}class Mb extends _c{async _call(S){return new Et(await super._call(S))}}class gc extends z{}class bb extends gc{}class yb extends gc{async _call(S){return new Et(await super._call(S))}}class si extends z{}class vb extends si{}class xb extends si{async _call(S){return new Et(await super._call(S))}}class Tb extends si{}class wc extends z{}class Eb extends wc{}class Pb extends wc{}class Mc extends z{}class Cb extends Mc{}class Sb extends Mc{}class $b extends z{}class Ab extends $b{}class ni extends z{}class kb extends ni{}class Ib extends ni{}class Fb extends ni{}class Ob extends z{}class Db extends Ob{}class Lb extends z{}class zb extends Lb{}class Rb extends z{}class Bb extends Rb{}class bc extends z{}class Nb extends bc{}class jb extends bc{}class yc extends z{}class Vb extends yc{}class Ub extends yc{}class Wb extends z{}class Gb extends Wb{}class vc extends z{}class Hb extends vc{}class Kb extends vc{async _call(S){return new Et(await super._call(S))}}class xc extends z{}class qb extends xc{}class Qb extends xc{async _call(S){return new Et(await super._call(S))}}class Tc extends z{}class Xb extends Tc{}class Jb extends Tc{async _call(S){return new Et(await super._call(S))}}class Ec extends z{}class Yb extends Ec{}class Zb extends Ec{async _call(S){return new Et(await super._call(S))}}class ey extends z{}class ty extends ey{}class ry extends z{}class sy extends ry{}class ny extends z{}class oy extends ny{}class Pc extends z{}class ay extends Pc{}class iy extends Pc{async _call(S){return new ly(await super._call(S))}}class ly extends _e{constructor({logits:S,pred_boxes:L}){super(),this.logits=S,this.pred_boxes=L}}class uy extends z{}class cy extends uy{async get_image_embeddings({pixel_values:S}){return await Z(this,{pixel_values:S})}async forward(S){if((!S.image_embeddings||!S.image_positional_embeddings)&&(S={...S,...await this.get_image_embeddings(S)}),!S.input_labels&&S.input_points){const le=S.input_points.dims.slice(0,-1),ve=le.reduce((be,Fe)=>be*Fe,1);S.input_labels=new u.Tensor("int64",new BigInt64Array(ve).fill(1n),le)}const L={image_embeddings:S.image_embeddings,image_positional_embeddings:S.image_positional_embeddings};return S.input_points&&(L.input_points=S.input_points),S.input_labels&&(L.input_labels=S.input_labels),S.input_boxes&&(L.input_boxes=S.input_boxes),await U(this.sessions.prompt_encoder_mask_decoder,L)}async _call(S){return new dy(await super._call(S))}}class dy extends _e{constructor({iou_scores:S,pred_masks:L}){super(),this.iou_scores=S,this.pred_masks=L}}class Cc extends z{}class py extends Cc{}class hy extends Cc{}class Sc extends z{}class my extends Sc{}class fy extends Sc{}class un extends z{}class _y extends un{}class gy extends un{async _call(S){return new cn(await super._call(S))}}class wy extends un{async _call(S){return new Et(await super._call(S))}}class My extends un{async _call(S){return new Er(await super._call(S))}}class by extends z{}class yy extends by{async _call(S){return new cn(await super._call(S))}}class $c extends z{}class vy extends $c{}class xy extends $c{async _call(S){return new Er(await super._call(S))}}class Ty extends z{}class Ey extends Ty{}class oi extends z{}class Py extends oi{}class Cy extends oi{async _call(S){return new cn(await super._call(S))}}class Sy extends oi{async _call(S){return new Et(await super._call(S))}}class ea extends z{}class $y extends ea{}class Ay extends ea{async _call(S){return new cn(await super._call(S))}}class ky extends ea{async _call(S){return new Et(await super._call(S))}}class Iy extends ea{async _call(S){return new Er(await super._call(S))}}class ai extends z{}class Fy extends ai{}class Oy extends ai{async _call(S){return new cn(await super._call(S))}}class Dy extends ai{async _call(S){return new Et(await super._call(S))}}class wE extends z{}class Ly extends un{}class zy extends un{async _call(S){return new cn(await super._call(S))}}class Ry extends un{async _call(S){return new Et(await super._call(S))}}class io extends z{}class By extends io{}class Ny extends io{async _call(S){return new cn(await super._call(S))}}class jy extends io{async _call(S){return new Et(await super._call(S))}}class Vy extends io{async _call(S){return new tx(await super._call(S))}}class Uy extends io{async _call(S){return new Er(await super._call(S))}}class Wy extends z{}class Gy extends Wy{}class ii extends z{}class ME extends ii{}class Hy extends ii{}class Ky extends ii{async generate_speech(S,L,{threshold:le=.5,minlenratio:ve=0,maxlenratio:be=20,vocoder:Fe=null}={}){const Be={input_ids:S},{encoder_outputs:Ke,encoder_attention_mask:Ye}=await Z(this,Be),mt=Ke.dims[1]/this.config.reduction_factor,yt=Math.floor(mt*be),vt=Math.floor(mt*ve),Ot=this.config.num_mel_bins;let _t=[],$t=null,ct=null,Ft=0;for(;;){++Ft;const hr=q(!!ct);let lr;ct?lr=ct.output_sequence_out:lr=new u.Tensor("float32",new Float32Array(Ot),[1,1,Ot]);let br={use_cache_branch:hr,output_sequence:lr,encoder_attention_mask:Ye,speaker_embeddings:L,encoder_hidden_states:Ke};this.addPastKeyValues(br,$t),ct=await U(this.sessions.decoder_model_merged,br),$t=this.getPastKeyValues(ct,$t);const{prob:Lr,spectrum:At}=ct;if(_t.push(At),Ft>=vt&&(Array.from(Lr.data).filter(Pr=>Pr>=le).length>0||Ft>=yt))break}const Ht=(0,u.cat)(_t),{waveform:tr}=await U(Fe.sessions.model,{spectrogram:Ht});return{spectrogram:Ht,waveform:tr}}}class qy extends z{constructor(){super(...arguments);re(this,"main_input_name","spectrogram")}}class Qy extends z{}class Xy extends Qy{}class Ac extends z{}class Jy extends Ac{}class Yy extends Ac{}class kc extends z{}class Zy extends kc{}class e0 extends kc{}class Ic extends z{}class t0 extends Ic{}class r0 extends Ic{}class Fc extends z{}class s0 extends Fc{}class n0 extends Fc{}class li extends z{}class o0 extends li{}class a0 extends li{static async from_pretrained(S,L={}){return super.from_pretrained(S,{...L,model_file_name:L.model_file_name??"text_model"})}}class i0 extends li{static async from_pretrained(S,L={}){return super.from_pretrained(S,{...L,model_file_name:L.model_file_name??"audio_model"})}}class l0 extends z{}class Oc extends l0{async _call(S){return new sx(await super._call(S))}}class ui extends z{}class bE extends ui{}class u0 extends ui{}class c0 extends ui{}class Dc extends z{}class d0 extends Dc{}class p0 extends Dc{}class Lc extends z{}class h0 extends Lc{}class m0 extends Lc{async _call(S){return new Et(await super._call(S))}}class zc extends z{}class yE extends zc{}class vE extends zc{}class Rc extends z{constructor(){super(...arguments);re(this,"forward_params",["input_ids","attention_mask","encoder_outputs","decoder_input_ids","decoder_attention_mask","past_key_values"])}_apply_and_filter_by_delay_pattern_mask(L){const[le,ve]=L.dims,be=this.config.decoder.num_codebooks,Fe=ve-be;let Be=0;for(let mt=0;mt<L.size;++mt){if(L.data[mt]===this.config.decoder.pad_token_id)continue;const yt=mt%ve,vt=Math.floor(mt/ve)%be,Ot=yt-vt;Ot>0&&Ot<=Fe&&(L.data[Be++]=L.data[mt])}const Ke=Math.floor(le/be),Ye=Be/(Ke*be);return new u.Tensor(L.type,L.data.slice(0,Be),[Ke,be,Ye])}prepare_inputs_for_generation(L,le,ve){let be=structuredClone(L);for(let Be=0;Be<be.length;++Be)for(let Ke=0;Ke<be[Be].length;++Ke)Be%this.config.decoder.num_codebooks>=Ke&&(be[Be][Ke]=BigInt(this.config.decoder.pad_token_id));return ve.guidance_scale!==null&&ve.guidance_scale>1&&(be=be.concat(be)),super.prepare_inputs_for_generation(be,le,ve)}async generate(L){const le=await super.generate(L),ve=this._apply_and_filter_by_delay_pattern_mask(le).unsqueeze_(0),{audio_values:be}=await U(this.sessions.encodec_decode,{audio_codes:ve});return be}}class ci extends z{}class f0 extends ci{}class _0 extends ci{async _call(S){return new Et(await super._call(S))}}class g0 extends ci{}class di extends z{}class w0 extends di{}class M0 extends di{async _call(S){return new Et(await super._call(S))}}class b0 extends di{}class pi extends z{}class y0 extends pi{}class v0 extends pi{async _call(S){return new Et(await super._call(S))}}class x0 extends pi{}class hi extends z{}class T0 extends hi{}class E0 extends hi{async _call(S){return new Et(await super._call(S))}}class P0 extends hi{}class C0 extends z{}class S0 extends C0{}class $0 extends z{}class A0 extends $0{constructor(...L){super(...L);re(this,"forward_params",["input_ids","pixel_values","images_seq_mask","images_emb_mask","attention_mask","position_ids","past_key_values"]);this._generation_mode="text"}async forward(L){const le=this._generation_mode??"text";let ve;if(le==="text"||!L.past_key_values){const Ye=this.sessions.prepare_inputs_embeds,mt=(0,i.pick)(L,Ye.inputNames);ve=await U(Ye,mt)}else{const Ye=this.sessions.gen_img_embeds,mt=(0,i.pick)({image_ids:L.input_ids},Ye.inputNames);ve=await U(Ye,mt)}const be={...L,...ve},Fe=await he(this,be),Be=this.sessions[le==="text"?"lm_head":"gen_head"];if(!Be)throw new Error(`Unable to find "${Be}" generation head`);const Ke=await U(Be,(0,i.pick)(Fe,Be.inputNames));return{...ve,...Fe,...Ke}}async generate(L){return this._generation_mode="text",super.generate(L)}async generate_images(L){this._generation_mode="image";const le=(L.inputs??L[this.main_input_name]).dims[1],be=(await super.generate(L)).slice(null,[le,null]),Fe=this.sessions.image_decode,{decoded_image:Be}=await U(Fe,{generated_tokens:be}),Ke=Be.add_(1).mul_(255/2).clamp_(0,255).to("uint8"),Ye=[];for(const mt of Ke){const yt=_.RawImage.fromTensor(mt);Ye.push(yt)}return Ye}}class k0 extends _e{constructor({char_logits:S,bpe_logits:L,wp_logits:le}){super(),this.char_logits=S,this.bpe_logits=L,this.wp_logits=le}get logits(){return[this.char_logits,this.bpe_logits,this.wp_logits]}}class I0 extends z{}class F0 extends I0{async _call(S){return new k0(await super._call(S))}}class Bc extends z{}class O0 extends Bc{}class D0 extends Bc{}class Nc extends z{}class L0 extends Nc{}class z0 extends Nc{}class R0 extends z{constructor(){super(...arguments);re(this,"forward_params",["input_ids","attention_mask","position_ids","audio_values","past_key_values"])}}class jc extends R0{_merge_input_ids_with_audio_features(S){const L=S.audio_features.dims.at(-1),le=S.audio_features.view(-1,L);return O({audio_token_id:this.config.ignore_index??this.config.audio_token_id,...S,audio_features:le})}}class B0 extends jc{}class mi extends z{constructor(){super(...arguments);re(this,"main_input_name","input_values");re(this,"forward_params",["input_values"])}}class N0 extends _e{constructor({audio_codes:S}){super(),this.audio_codes=S}}class j0 extends _e{constructor({audio_values:S}){super(),this.audio_values=S}}class V0 extends mi{async encode(S){return new N0(await U(this.sessions.encoder_model,S))}async decode(S){return new j0(await U(this.sessions.decoder_model,S))}}class U0 extends mi{static async from_pretrained(S,L={}){return super.from_pretrained(S,{...L,model_file_name:L.model_file_name??"encoder_model"})}}class W0 extends mi{static async from_pretrained(S,L={}){return super.from_pretrained(S,{...L,model_file_name:L.model_file_name??"decoder_model"})}}class fi extends z{constructor(){super(...arguments);re(this,"main_input_name","input_values");re(this,"forward_params",["input_values"])}}class G0 extends _e{constructor({audio_codes:S}){super(),this.audio_codes=S}}class H0 extends _e{constructor({audio_values:S}){super(),this.audio_values=S}}class K0 extends fi{async encode(S){return new G0(await U(this.sessions.encoder_model,S))}async decode(S){return new H0(await U(this.sessions.decoder_model,S))}}class q0 extends fi{static async from_pretrained(S,L={}){return super.from_pretrained(S,{...L,model_file_name:L.model_file_name??"encoder_model"})}}class Q0 extends fi{static async from_pretrained(S,L={}){return super.from_pretrained(S,{...L,model_file_name:L.model_file_name??"decoder_model"})}}class _i extends z{constructor(){super(...arguments);re(this,"main_input_name","input_values");re(this,"forward_params",["input_values"])}}class X0 extends _i{async encode(S){return await U(this.sessions.encoder_model,S)}async decode(S){return await U(this.sessions.decoder_model,S)}}class J0 extends _i{static async from_pretrained(S,L={}){return super.from_pretrained(S,{...L,model_file_name:L.model_file_name??"encoder_model"})}}class Y0 extends _i{static async from_pretrained(S,L={}){return super.from_pretrained(S,{...L,model_file_name:L.model_file_name??"decoder_model"})}}class Nt{static async from_pretrained(S,{progress_callback:L=null,config:le=null,cache_dir:ve=null,local_files_only:be=!1,revision:Fe="main",model_file_name:Be=null,subfolder:Ke="onnx",device:Ye=null,dtype:mt=null,use_external_data_format:yt=null,session_options:vt={}}={}){const Ot={progress_callback:L,config:le,cache_dir:ve,local_files_only:be,revision:Fe,model_file_name:Be,subfolder:Ke,device:Ye,dtype:mt,use_external_data_format:yt,session_options:vt};if(Ot.config=await s.AutoConfig.from_pretrained(S,Ot),!this.MODEL_CLASS_MAPPINGS)throw new Error("`MODEL_CLASS_MAPPINGS` not implemented for this type of `AutoClass`: "+this.name);const _t=Ot.config.model_type;for(const $t of this.MODEL_CLASS_MAPPINGS){let ct=$t.get(_t);if(!ct){for(const Ft of $t.values())if(Ft[0]===_t){ct=Ft;break}if(!ct)continue}return await ct[1].from_pretrained(S,Ot)}if(this.BASE_IF_FAIL)return Tv.has(_t)||console.warn(`Unknown model class "${_t}", attempting to construct from base class.`),await z.from_pretrained(S,Ot);throw Error(`Unsupported model type: ${_t}`)}}re(Nt,"MODEL_CLASS_MAPPINGS",null),re(Nt,"BASE_IF_FAIL",!1);const xE=new Map([["bert",["BertModel",Ce]],["neobert",["NeoBertModel",Le]],["modernbert",["ModernBertModel",ze]],["nomic_bert",["NomicBertModel",Vr]],["roformer",["RoFormerModel",sr]],["electra",["ElectraModel",ee]],["esm",["EsmModel",fs]],["convbert",["ConvBertModel",ft]],["camembert",["CamembertModel",nt]],["deberta",["DebertaModel",qt]],["deberta-v2",["DebertaV2Model",ps]],["mpnet",["MPNetModel",Tn]],["albert",["AlbertModel",Ge]],["distilbert",["DistilBertModel",ms]],["roberta",["RobertaModel",Ca]],["xlm",["XLMModel",ka]],["xlm-roberta",["XLMRobertaModel",La]],["clap",["ClapModel",o0]],["clip",["CLIPModel",Zn]],["clipseg",["CLIPSegModel",os]],["chinese_clip",["ChineseCLIPModel",kn]],["siglip",["SiglipModel",Bo]],["jina_clip",["JinaCLIPModel",ti]],["mobilebert",["MobileBertModel",rt]],["squeezebert",["SqueezeBertModel",G]],["wav2vec2",["Wav2Vec2Model",_y]],["wav2vec2-bert",["Wav2Vec2BertModel",Fy]],["unispeech",["UniSpeechModel",Py]],["unispeech-sat",["UniSpeechSatModel",$y]],["hubert",["HubertModel",Ly]],["wavlm",["WavLMModel",By]],["audio-spectrogram-transformer",["ASTModel",Ba]],["vits",["VitsModel",Oc]],["pyannote",["PyAnnoteModel",vy]],["wespeaker-resnet",["WeSpeakerResNetModel",Ey]],["detr",["DetrModel",eb]],["rt_detr",["RTDetrModel",sb]],["rt_detr_v2",["RTDetrV2Model",ob]],["rf_detr",["RFDetrModel",lb]],["d_fine",["DFineModel",db]],["table-transformer",["TableTransformerModel",hb]],["vit",["ViTModel",PM]],["ijepa",["IJepaModel",SM]],["pvt",["PvtModel",IM]],["vit_msn",["ViTMSNModel",LM]],["vit_mae",["ViTMAEModel",DM]],["groupvit",["GroupViTModel",BM]],["fastvit",["FastViTModel",NM]],["mobilevit",["MobileViTModel",WM]],["mobilevitv2",["MobileViTV2Model",HM]],["owlvit",["OwlViTModel",qM]],["owlv2",["Owlv2Model",XM]],["beit",["BeitModel",YM]],["deit",["DeiTModel",_b]],["hiera",["HieraModel",wb]],["convnext",["ConvNextModel",Hb]],["convnextv2",["ConvNextV2Model",qb]],["dinov2",["Dinov2Model",Xb]],["dinov2_with_registers",["Dinov2WithRegistersModel",Yb]],["dinov3_vit",["DINOv3ViTModel",ty]],["dinov3_convnext",["DINOv3ConvNextModel",sy]],["resnet",["ResNetModel",bb]],["swin",["SwinModel",vb]],["swin2sr",["Swin2SRModel",Eb]],["donut-swin",["DonutSwinModel",Gb]],["yolos",["YolosModel",ay]],["dpt",["DPTModel",Cb]],["glpn",["GLPNModel",Vb]],["hifigan",["SpeechT5HifiGan",qy]],["efficientnet",["EfficientNetModel",h0]],["decision_transformer",["DecisionTransformerModel",S0]],["patchtst",["PatchTSTForPrediction",O0]],["patchtsmixer",["PatchTSMixerForPrediction",L0]],["mobilenet_v1",["MobileNetV1Model",f0]],["mobilenet_v2",["MobileNetV2Model",w0]],["mobilenet_v3",["MobileNetV3Model",y0]],["mobilenet_v4",["MobileNetV4Model",T0]],["maskformer",["MaskFormerModel",Nb]],["mgp-str",["MgpstrForSceneTextRecognition",F0]],["style_text_to_speech_2",["StyleTextToSpeech2Model",Gy]]]),TE=new Map([["t5",["T5Model",Rt]],["longt5",["LongT5Model",Gt]],["mt5",["MT5Model",dr]],["bart",["BartModel",ns]],["mbart",["MBartModel",Zr]],["marian",["MarianModel",py]],["whisper",["WhisperModel",Fo]],["m2m_100",["M2M100Model",my]],["blenderbot",["BlenderbotModel",pr]],["blenderbot-small",["BlenderbotSmallModel",Qr]]]),EE=new Map([["mimi",["MimiModel",V0]],["dac",["DacModel",K0]],["snac",["SnacModel",X0]]]),PE=new Map([["bloom",["BloomModel",bM]],["jais",["JAISModel",ro]],["gpt2",["GPT2Model",Wo]],["gptj",["GPTJModel",Qo]],["gpt_bigcode",["GPTBigCodeModel",Jo]],["gpt_neo",["GPTNeoModel",Ko]],["gpt_neox",["GPTNeoXModel",no]],["codegen",["CodeGenModel",I]],["llama",["LlamaModel",pe]],["nanochat",["NanoChatModel",Tt]],["arcee",["ArceeModel",Tr]],["lfm2",["Lfm2Model",Iw]],["smollm3",["SmolLM3Model",Ow]],["exaone",["ExaoneModel",Nw]],["olmo",["OlmoModel",Ww]],["olmo2",["Olmo2Model",Hw]],["mobilellm",["MobileLLMModel",Vw]],["granite",["GraniteModel",qw]],["granitemoehybrid",["GraniteMoeHybridModel",Xw]],["cohere",["CohereModel",Yw]],["gemma",["GemmaModel",eM]],["gemma2",["Gemma2Model",rM]],["vaultgemma",["VaultGemmaModel",nM]],["gemma3_text",["Gemma3Model",aM]],["helium",["HeliumModel",Lw]],["glm",["GlmModel",Rw]],["openelm",["OpenELMModel",lM]],["qwen2",["Qwen2Model",cM]],["qwen3",["Qwen3Model",pM]],["phi",["PhiModel",_M]],["phi3",["Phi3Model",wM]],["mpt",["MptModel",vM]],["opt",["OPTModel",TM]],["mistral",["MistralModel",Jy]],["ernie4_5",["Ernie4_5_Model",Zy]],["starcoder2",["Starcoder2Model",t0]],["falcon",["FalconModel",s0]],["stablelm",["StableLmModel",d0]],["modernbert-decoder",["ModernBertDecoderModel",ht]]]),Vc=new Map([["speecht5",["SpeechT5ForSpeechToText",Hy]],["whisper",["WhisperForConditionalGeneration",qs]],["lite-whisper",["LiteWhisperForConditionalGeneration",ja]],["moonshine",["MoonshineForConditionalGeneration",Va]]]),Z0=new Map([["speecht5",["SpeechT5ForTextToSpeech",Ky]]]),ev=new Map([["vits",["VitsModel",Oc]],["musicgen",["MusicgenForConditionalGeneration",Rc]]]),tv=new Map([["bert",["BertForSequenceClassification",de]],["neobert",["NeoBertForSequenceClassification",We]],["modernbert",["ModernBertForSequenceClassification",gt]],["roformer",["RoFormerForSequenceClassification",rn]],["electra",["ElectraForSequenceClassification",Y]],["esm",["EsmForSequenceClassification",Gr]],["convbert",["ConvBertForSequenceClassification",Ds]],["camembert",["CamembertForSequenceClassification",pt]],["deberta",["DebertaForSequenceClassification",qr]],["deberta-v2",["DebertaV2ForSequenceClassification",Ir]],["mpnet",["MPNetForSequenceClassification",Pn]],["albert",["AlbertForSequenceClassification",Xe]],["distilbert",["DistilBertForSequenceClassification",Yr]],["roberta",["RobertaForSequenceClassification",Ao]],["xlm",["XLMForSequenceClassification",Fa]],["xlm-roberta",["XLMRobertaForSequenceClassification",Ra]],["bart",["BartForSequenceClassification",Mr]],["mbart",["MBartForSequenceClassification",Or]],["mobilebert",["MobileBertForSequenceClassification",Hs]],["squeezebert",["SqueezeBertForSequenceClassification",ue]]]),rv=new Map([["bert",["BertForTokenClassification",we]],["neobert",["NeoBertForTokenClassification",qe]],["modernbert",["ModernBertForTokenClassification",dt]],["roformer",["RoFormerForTokenClassification",sn]],["electra",["ElectraForTokenClassification",oe]],["esm",["EsmForTokenClassification",Ne]],["convbert",["ConvBertForTokenClassification",St]],["camembert",["CamembertForTokenClassification",xt]],["deberta",["DebertaForTokenClassification",nr]],["deberta-v2",["DebertaV2ForTokenClassification",Ls]],["mpnet",["MPNetForTokenClassification",Cn]],["distilbert",["DistilBertForTokenClassification",Rr]],["roberta",["RobertaForTokenClassification",$a]],["xlm",["XLMForTokenClassification",Oa]],["xlm-roberta",["XLMRobertaForTokenClassification",ko]]]),Uc=new Map([["t5",["T5ForConditionalGeneration",Lt]],["longt5",["LongT5ForConditionalGeneration",_r]],["mt5",["MT5ForConditionalGeneration",wr]],["bart",["BartForConditionalGeneration",Yt]],["mbart",["MBartForConditionalGeneration",$s]],["marian",["MarianMTModel",hy]],["m2m_100",["M2M100ForConditionalGeneration",fy]],["blenderbot",["BlenderbotForConditionalGeneration",fr]],["blenderbot-small",["BlenderbotSmallForConditionalGeneration",Ks]]]),Wc=new Map([["bloom",["BloomForCausalLM",yM]],["gpt2",["GPT2LMHeadModel",Go]],["jais",["JAISLMHeadModel",Ho]],["gptj",["GPTJForCausalLM",Xo]],["gpt_bigcode",["GPTBigCodeForCausalLM",h]],["gpt_neo",["GPTNeoForCausalLM",On]],["gpt_neox",["GPTNeoXForCausalLM",qo]],["codegen",["CodeGenForCausalLM",R]],["llama",["LlamaForCausalLM",Ie]],["nanochat",["NanoChatForCausalLM",zt]],["llama4_text",["Llama4ForCausalLM",Je]],["arcee",["ArceeForCausalLM",Qs]],["lfm2",["Lfm2ForCausalLM",Fw]],["smollm3",["SmolLM3ForCausalLM",Dw]],["exaone",["ExaoneForCausalLM",jw]],["olmo",["OlmoForCausalLM",Gw]],["olmo2",["Olmo2ForCausalLM",Kw]],["mobilellm",["MobileLLMForCausalLM",Uw]],["granite",["GraniteForCausalLM",Qw]],["granitemoehybrid",["GraniteMoeHybridForCausalLM",Jw]],["cohere",["CohereForCausalLM",Zw]],["gemma",["GemmaForCausalLM",tM]],["gemma2",["Gemma2ForCausalLM",sM]],["vaultgemma",["VaultGemmaForCausalLM",oM]],["gemma3_text",["Gemma3ForCausalLM",iM]],["helium",["HeliumForCausalLM",zw]],["glm",["GlmForCausalLM",Bw]],["openelm",["OpenELMForCausalLM",uM]],["qwen2",["Qwen2ForCausalLM",dM]],["qwen3",["Qwen3ForCausalLM",hM]],["phi",["PhiForCausalLM",gM]],["phi3",["Phi3ForCausalLM",MM]],["mpt",["MptForCausalLM",xM]],["opt",["OPTForCausalLM",EM]],["mbart",["MBartForCausalLM",_s]],["mistral",["MistralForCausalLM",Yy]],["ernie4_5",["Ernie4_5_ForCausalLM",e0]],["starcoder2",["Starcoder2ForCausalLM",r0]],["falcon",["FalconForCausalLM",n0]],["trocr",["TrOCRForCausalLM",Xy]],["stablelm",["StableLmForCausalLM",p0]],["modernbert-decoder",["ModernBertDecoderForCausalLM",yr]],["phi3_v",["Phi3VForCausalLM",Lo]]]),CE=new Map([["multi_modality",["MultiModalityCausalLM",A0]]]),sv=new Map([["bert",["BertForMaskedLM",ye]],["neobert",["NeoBertForMaskedLM",Te]],["modernbert",["ModernBertForMaskedLM",He]],["roformer",["RoFormerForMaskedLM",Ar]],["electra",["ElectraForMaskedLM",V]],["esm",["EsmForMaskedLM",Gs]],["convbert",["ConvBertForMaskedLM",Os]],["camembert",["CamembertForMaskedLM",wt]],["deberta",["DebertaForMaskedLM",Wr]],["deberta-v2",["DebertaV2ForMaskedLM",hs]],["mpnet",["MPNetForMaskedLM",En]],["albert",["AlbertForMaskedLM",et]],["distilbert",["DistilBertForMaskedLM",mr]],["roberta",["RobertaForMaskedLM",Sa]],["xlm",["XLMWithLMHeadModel",Ia]],["xlm-roberta",["XLMRobertaForMaskedLM",za]],["mobilebert",["MobileBertForMaskedLM",Qt]],["squeezebert",["SqueezeBertForMaskedLM",se]]]),nv=new Map([["bert",["BertForQuestionAnswering",ce]],["neobert",["NeoBertForQuestionAnswering",st]],["roformer",["RoFormerForQuestionAnswering",nn]],["electra",["ElectraForQuestionAnswering",xe]],["convbert",["ConvBertForQuestionAnswering",Kt]],["camembert",["CamembertForQuestionAnswering",tt]],["deberta",["DebertaForQuestionAnswering",kr]],["deberta-v2",["DebertaV2ForQuestionAnswering",zs]],["mpnet",["MPNetForQuestionAnswering",ge]],["albert",["AlbertForQuestionAnswering",Qe]],["distilbert",["DistilBertForQuestionAnswering",Cs]],["roberta",["RobertaForQuestionAnswering",Aa]],["xlm",["XLMForQuestionAnswering",Da]],["xlm-roberta",["XLMRobertaForQuestionAnswering",Io]],["mobilebert",["MobileBertForQuestionAnswering",Ss]],["squeezebert",["SqueezeBertForQuestionAnswering",fe]]]),Gc=new Map([["vision-encoder-decoder",["VisionEncoderDecoderModel",Oo]],["idefics3",["Idefics3ForConditionalGeneration",Bs]],["smolvlm",["SmolVLMForConditionalGeneration",Yn]]]),ov=new Map([["llava",["LlavaForConditionalGeneration",$n]],["llava_onevision",["LlavaOnevisionForConditionalGeneration",Ua]],["moondream1",["Moondream1ForConditionalGeneration",Wa]],["florence2",["Florence2ForConditionalGeneration",Ha]],["qwen2-vl",["Qwen2VLForConditionalGeneration",fM]],["idefics3",["Idefics3ForConditionalGeneration",Bs]],["smolvlm",["SmolVLMForConditionalGeneration",Yn]],["paligemma",["PaliGemmaForConditionalGeneration",qa]],["llava_qwen2",["LlavaQwen2ForCausalLM",Qa]],["gemma3n",["Gemma3nForConditionalGeneration",Do]]]),av=new Map([["ultravox",["UltravoxModel",jc]],["voxtral",["VoxtralForConditionalGeneration",B0]]]),SE=new Map([["vision-encoder-decoder",["VisionEncoderDecoderModel",Oo]]]),iv=new Map([["vit",["ViTForImageClassification",CM]],["ijepa",["IJepaForImageClassification",$M]],["pvt",["PvtForImageClassification",FM]],["vit_msn",["ViTMSNForImageClassification",zM]],["fastvit",["FastViTForImageClassification",jM]],["mobilevit",["MobileViTForImageClassification",GM]],["mobilevitv2",["MobileViTV2ForImageClassification",KM]],["beit",["BeitForImageClassification",ZM]],["deit",["DeiTForImageClassification",gb]],["hiera",["HieraForImageClassification",Mb]],["convnext",["ConvNextForImageClassification",Kb]],["convnextv2",["ConvNextV2ForImageClassification",Qb]],["dinov2",["Dinov2ForImageClassification",Jb]],["dinov2_with_registers",["Dinov2WithRegistersForImageClassification",Zb]],["resnet",["ResNetForImageClassification",yb]],["swin",["SwinForImageClassification",xb]],["segformer",["SegformerForImageClassification",u0]],["efficientnet",["EfficientNetForImageClassification",m0]],["mobilenet_v1",["MobileNetV1ForImageClassification",_0]],["mobilenet_v2",["MobileNetV2ForImageClassification",M0]],["mobilenet_v3",["MobileNetV3ForImageClassification",v0]],["mobilenet_v4",["MobileNetV4ForImageClassification",E0]]]),lv=new Map([["detr",["DetrForObjectDetection",tb]],["rt_detr",["RTDetrForObjectDetection",nb]],["rt_detr_v2",["RTDetrV2ForObjectDetection",ab]],["rf_detr",["RFDetrForObjectDetection",ub]],["d_fine",["DFineForObjectDetection",pb]],["table-transformer",["TableTransformerForObjectDetection",mb]],["yolos",["YolosForObjectDetection",iy]]]),uv=new Map([["owlvit",["OwlViTForObjectDetection",QM]],["owlv2",["Owlv2ForObjectDetection",JM]],["grounding-dino",["GroundingDinoForObjectDetection",oy]]]),lo=new Map([["detr",["DetrForSegmentation",lc]],["clipseg",["CLIPSegForImageSegmentation",In]]]),cv=new Map([["segformer",["SegformerForSemanticSegmentation",c0]],["sapiens",["SapiensForSemanticSegmentation",kb]],["swin",["SwinForSemanticSegmentation",Tb]],["mobilenet_v1",["MobileNetV1ForSemanticSegmentation",g0]],["mobilenet_v2",["MobileNetV2ForSemanticSegmentation",b0]],["mobilenet_v3",["MobileNetV3ForSemanticSegmentation",x0]],["mobilenet_v4",["MobileNetV4ForSemanticSegmentation",P0]]]),dv=new Map([["detr",["DetrForSegmentation",lc]],["maskformer",["MaskFormerForInstanceSegmentation",jb]]]),pv=new Map([["sam",["SamModel",cy]]]),hv=new Map([["wav2vec2",["Wav2Vec2ForCTC",gy]],["wav2vec2-bert",["Wav2Vec2BertForCTC",Oy]],["unispeech",["UniSpeechForCTC",Cy]],["unispeech-sat",["UniSpeechSatForCTC",Ay]],["wavlm",["WavLMForCTC",Ny]],["hubert",["HubertForCTC",zy]],["parakeet_ctc",["ParakeetForCTC",yy]]]),mv=new Map([["wav2vec2",["Wav2Vec2ForSequenceClassification",wy]],["wav2vec2-bert",["Wav2Vec2BertForSequenceClassification",Dy]],["unispeech",["UniSpeechForSequenceClassification",Sy]],["unispeech-sat",["UniSpeechSatForSequenceClassification",ky]],["wavlm",["WavLMForSequenceClassification",jy]],["hubert",["HubertForSequenceClassification",Ry]],["audio-spectrogram-transformer",["ASTForAudioClassification",Na]]]),fv=new Map([["wavlm",["WavLMForXVector",Vy]]]),_v=new Map([["unispeech-sat",["UniSpeechSatForAudioFrameClassification",Iy]],["wavlm",["WavLMForAudioFrameClassification",Uy]],["wav2vec2",["Wav2Vec2ForAudioFrameClassification",My]],["pyannote",["PyAnnoteForAudioFrameClassification",xy]]]),gv=new Map([["vitmatte",["VitMatteForImageMatting",UM]]]),$E=new Map([["patchtst",["PatchTSTForPrediction",D0]],["patchtsmixer",["PatchTSMixerForPrediction",z0]]]),wv=new Map([["swin2sr",["Swin2SRForImageSuperResolution",Pb]]]),Mv=new Map([["dpt",["DPTForDepthEstimation",Sb]],["depth_anything",["DepthAnythingForDepthEstimation",Ab]],["glpn",["GLPNForDepthEstimation",Ub]],["sapiens",["SapiensForDepthEstimation",Ib]],["depth_pro",["DepthProForDepthEstimation",Db]],["metric3d",["Metric3DForDepthEstimation",zb]],["metric3dv2",["Metric3Dv2ForDepthEstimation",Bb]]]),bv=new Map([["sapiens",["SapiensForNormalEstimation",Fb]]]),yv=new Map([["vitpose",["VitPoseForPoseEstimation",kM]]]),vv=new Map([["clip",["CLIPVisionModelWithProjection",Ro]],["siglip",["SiglipVisionModel",No]],["jina_clip",["JinaCLIPVisionModel",Uo]]]),xv=[[xE,x.EncoderOnly],[TE,x.EncoderDecoder],[PE,x.DecoderOnly],[EE,x.AutoEncoder],[tv,x.EncoderOnly],[rv,x.EncoderOnly],[Uc,x.Seq2Seq],[Vc,x.Seq2Seq],[Wc,x.DecoderOnly],[CE,x.MultiModality],[sv,x.EncoderOnly],[nv,x.EncoderOnly],[Gc,x.Vision2Seq],[ov,x.ImageTextToText],[av,x.AudioTextToText],[iv,x.EncoderOnly],[lo,x.EncoderOnly],[dv,x.EncoderOnly],[cv,x.EncoderOnly],[gv,x.EncoderOnly],[$E,x.EncoderOnly],[wv,x.EncoderOnly],[Mv,x.EncoderOnly],[bv,x.EncoderOnly],[yv,x.EncoderOnly],[lv,x.EncoderOnly],[uv,x.EncoderOnly],[pv,x.MaskGeneration],[hv,x.EncoderOnly],[mv,x.EncoderOnly],[Z0,x.Seq2Seq],[ev,x.EncoderOnly],[fv,x.EncoderOnly],[_v,x.EncoderOnly],[vv,x.EncoderOnly]];for(const[w,S]of xv)for(const[L,le]of w.values())M.set(L,S),v.set(le,L),T.set(L,le);const AE=[["MusicgenForConditionalGeneration",Rc,x.Musicgen],["Phi3VForCausalLM",Lo,x.Phi3V],["CLIPTextModelWithProjection",zo,x.EncoderOnly],["SiglipTextModel",ei,x.EncoderOnly],["JinaCLIPTextModel",Vo,x.EncoderOnly],["ClapTextModelWithProjection",a0,x.EncoderOnly],["ClapAudioModelWithProjection",i0,x.EncoderOnly],["DacEncoderModel",q0,x.EncoderOnly],["DacDecoderModel",Q0,x.EncoderOnly],["MimiEncoderModel",U0,x.EncoderOnly],["MimiDecoderModel",W0,x.EncoderOnly],["SnacEncoderModel",J0,x.EncoderOnly],["SnacDecoderModel",Y0,x.EncoderOnly],["Gemma3nForConditionalGeneration",Do,x.ImageAudioTextToText]];for(const[w,S,L]of AE)M.set(w,L),v.set(S,w),T.set(w,S);const Tv=new Map([["modnet",lo],["birefnet",lo],["isnet",lo],["ben",lo]]);for(const[w,S]of Tv.entries())S.set(w,["PreTrainedModel",z]),M.set(w,x.EncoderOnly),v.set(z,w),T.set(w,z);class Hc extends Nt{}re(Hc,"MODEL_CLASS_MAPPINGS",xv.map(S=>S[0])),re(Hc,"BASE_IF_FAIL",!0);class Ev extends Nt{}re(Ev,"MODEL_CLASS_MAPPINGS",[tv]);class Pv extends Nt{}re(Pv,"MODEL_CLASS_MAPPINGS",[rv]);class Cv extends Nt{}re(Cv,"MODEL_CLASS_MAPPINGS",[Uc]);class Sv extends Nt{}re(Sv,"MODEL_CLASS_MAPPINGS",[Vc]);class $v extends Nt{}re($v,"MODEL_CLASS_MAPPINGS",[Z0]);class Av extends Nt{}re(Av,"MODEL_CLASS_MAPPINGS",[ev]);class kv extends Nt{}re(kv,"MODEL_CLASS_MAPPINGS",[Wc]);class Iv extends Nt{}re(Iv,"MODEL_CLASS_MAPPINGS",[sv]);class Fv extends Nt{}re(Fv,"MODEL_CLASS_MAPPINGS",[nv]);class Ov extends Nt{}re(Ov,"MODEL_CLASS_MAPPINGS",[Gc]);class Dv extends Nt{}re(Dv,"MODEL_CLASS_MAPPINGS",[iv]);class Lv extends Nt{}re(Lv,"MODEL_CLASS_MAPPINGS",[lo]);class zv extends Nt{}re(zv,"MODEL_CLASS_MAPPINGS",[cv]);class Rv extends Nt{}re(Rv,"MODEL_CLASS_MAPPINGS",[dv]);class Bv extends Nt{}re(Bv,"MODEL_CLASS_MAPPINGS",[lv]);class Nv extends Nt{}re(Nv,"MODEL_CLASS_MAPPINGS",[uv]);class jv extends Nt{}re(jv,"MODEL_CLASS_MAPPINGS",[pv]);class Vv extends Nt{}re(Vv,"MODEL_CLASS_MAPPINGS",[hv]);class Uv extends Nt{}re(Uv,"MODEL_CLASS_MAPPINGS",[mv]);class Wv extends Nt{}re(Wv,"MODEL_CLASS_MAPPINGS",[fv]);class Gv extends Nt{}re(Gv,"MODEL_CLASS_MAPPINGS",[_v]);class Hv extends Nt{}re(Hv,"MODEL_CLASS_MAPPINGS",[SE]);class Kv extends Nt{}re(Kv,"MODEL_CLASS_MAPPINGS",[gv]);class qv extends Nt{}re(qv,"MODEL_CLASS_MAPPINGS",[wv]);class Qv extends Nt{}re(Qv,"MODEL_CLASS_MAPPINGS",[Mv]);class Xv extends Nt{}re(Xv,"MODEL_CLASS_MAPPINGS",[bv]);class Jv extends Nt{}re(Jv,"MODEL_CLASS_MAPPINGS",[yv]);class Yv extends Nt{}re(Yv,"MODEL_CLASS_MAPPINGS",[vv]);class Zv extends Nt{}re(Zv,"MODEL_CLASS_MAPPINGS",[ov]);class ex extends Nt{}re(ex,"MODEL_CLASS_MAPPINGS",[av]);class kE extends _e{constructor({logits:S,past_key_values:L,encoder_outputs:le,decoder_attentions:ve=null,cross_attentions:be=null}){super(),this.logits=S,this.past_key_values=L,this.encoder_outputs=le,this.decoder_attentions=ve,this.cross_attentions=be}}class Et extends _e{constructor({logits:S,...L}){super(),this.logits=S;const le=Object.values(L);le.length>0&&(this.attentions=le)}}class tx extends _e{constructor({logits:S,embeddings:L}){super(),this.logits=S,this.embeddings=L}}class Er extends _e{constructor({logits:S}){super(),this.logits=S}}class Dr extends _e{constructor({logits:S}){super(),this.logits=S}}class Nr extends _e{constructor({start_logits:S,end_logits:L}){super(),this.start_logits=S,this.end_logits=L}}class cn extends _e{constructor({logits:S}){super(),this.logits=S}}class IE extends _e{constructor({logits:S,past_key_values:L}){super(),this.logits=S,this.past_key_values=L}}class rx extends _e{constructor({alphas:S}){super(),this.alphas=S}}class sx extends _e{constructor({waveform:S,spectrogram:L}){super(),this.waveform=S,this.spectrogram=L}}},"./src/models/audio_spectrogram_transformer/feature_extraction_audio_spectrogram_transformer.js":(e,r,t)=>{t.r(r),t.d(r,{ASTFeatureExtractor:()=>n});var s=t("./src/base/feature_extraction_utils.js");t("./src/utils/tensor.js");var o=t("./src/utils/audio.js");class n extends s.FeatureExtractor{constructor(i){super(i);const l=this.config.sampling_rate,c=(0,o.mel_filter_bank)(257,this.config.num_mel_bins,20,Math.floor(l/2),l,null,"kaldi",!0);this.mel_filters=c,this.window=(0,o.window_function)(400,"hann",{periodic:!1}),this.mean=this.config.mean,this.std=this.config.std}async _extract_fbank_features(i,l){return(0,o.spectrogram)(i,this.window,400,160,{fft_length:512,power:2,center:!1,preemphasis:.97,mel_filters:this.mel_filters,log_mel:"log",mel_floor:1192092955078125e-22,remove_dc_offset:!0,max_num_frames:l,transpose:!0})}async _call(i){(0,s.validate_audio_inputs)(i,"ASTFeatureExtractor");const l=await this._extract_fbank_features(i,this.config.max_length);if(this.config.do_normalize){const c=this.std*2,p=l.data;for(let d=0;d<p.length;++d)p[d]=(p[d]-this.mean)/c}return{input_values:l.unsqueeze_(0)}}}},"./src/models/auto/feature_extraction_auto.js":(e,r,t)=>{t.r(r),t.d(r,{AutoFeatureExtractor:()=>a});var s=t("./src/utils/constants.js"),o=t("./src/utils/hub.js");t("./src/base/feature_extraction_utils.js");var n=t("./src/models/feature_extractors.js");class a{static async from_pretrained(l,c={}){const p=await(0,o.getModelJSON)(l,s.FEATURE_EXTRACTOR_NAME,!0,c),d=p.feature_extractor_type,u=n[d];if(!u)throw new Error(`Unknown feature_extractor_type: '${d}'. Please report this at ${s.GITHUB_ISSUE_URL}.`);return new u(p)}}},"./src/models/auto/image_processing_auto.js":(e,r,t)=>{t.r(r),t.d(r,{AutoImageProcessor:()=>i});var s=t("./src/utils/constants.js"),o=t("./src/utils/hub.js"),n=t("./src/base/image_processors_utils.js"),a=t("./src/models/image_processors.js");class i{static async from_pretrained(c,p={}){const d=await(0,o.getModelJSON)(c,s.IMAGE_PROCESSOR_NAME,!0,p),u=d.image_processor_type??d.feature_extractor_type;let _=a[u==null?void 0:u.replace(/Fast$/,"")];return _||(u!==void 0&&console.warn(`Image processor type '${u}' not found, assuming base ImageProcessor. Please report this at ${s.GITHUB_ISSUE_URL}.`),_=n.ImageProcessor),new _(d)}}},"./src/models/auto/processing_auto.js":(e,r,t)=>{t.r(r),t.d(r,{AutoProcessor:()=>c});var s=t("./src/utils/constants.js"),o=t("./src/utils/hub.js"),n=t("./src/base/processing_utils.js"),a=t("./src/models/processors.js"),i=t("./src/models/image_processors.js"),l=t("./src/models/feature_extractors.js");class c{static async from_pretrained(d,u={}){const _=await(0,o.getModelJSON)(d,s.IMAGE_PROCESSOR_NAME,!0,u),{image_processor_type:f,feature_extractor_type:b,processor_class:A}=_;if(A&&a[A])return a[A].from_pretrained(d,u);if(!f&&!b)throw new Error("No `image_processor_type` or `feature_extractor_type` found in the config.");const g={};if(f){const C=i[f.replace(/Fast$/,"")];if(!C)throw new Error(`Unknown image_processor_type: '${f}'.`);g.image_processor=new C(_)}if(b){const C=i[b];if(C)g.image_processor=new C(_);else{const x=l[b];if(!x)throw new Error(`Unknown feature_extractor_type: '${b}'.`);g.feature_extractor=new x(_)}}const y={};return new n.Processor(y,g,null)}}},"./src/models/beit/image_processing_beit.js":(e,r,t)=>{t.r(r),t.d(r,{BeitFeatureExtractor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{}},"./src/models/bit/image_processing_bit.js":(e,r,t)=>{t.r(r),t.d(r,{BitImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{}},"./src/models/chinese_clip/image_processing_chinese_clip.js":(e,r,t)=>{t.r(r),t.d(r,{ChineseCLIPFeatureExtractor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{}},"./src/models/clap/feature_extraction_clap.js":(e,r,t)=>{t.r(r),t.d(r,{ClapFeatureExtractor:()=>n});var s=t("./src/base/feature_extraction_utils.js");t("./src/utils/tensor.js");var o=t("./src/utils/audio.js");class n extends s.FeatureExtractor{constructor(i){super(i),this.mel_filters=(0,o.mel_filter_bank)(this.config.nb_frequency_bins,this.config.feature_size,this.config.frequency_min,this.config.frequency_max,this.config.sampling_rate,null,"htk"),this.mel_filters_slaney=(0,o.mel_filter_bank)(this.config.nb_frequency_bins,this.config.feature_size,this.config.frequency_min,this.config.frequency_max,this.config.sampling_rate,"slaney","slaney"),this.window=(0,o.window_function)(this.config.fft_window_size,"hann")}async _get_input_mel(i,l,c,p){let d;const u=i.length-l;if(u>0)if(c==="rand_trunc"){const _=Math.floor(Math.random()*(u+1));i=i.subarray(_,_+l),d=await this._extract_fbank_features(i,this.mel_filters_slaney,this.config.nb_max_samples)}else throw new Error(`Truncation strategy "${c}" not implemented`);else{if(u<0){let _=new Float64Array(l);if(_.set(i),p==="repeat")for(let f=i.length;f<l;f+=i.length)_.set(i.subarray(0,Math.min(i.length,l-f)),f);else if(p==="repeatpad")for(let f=i.length;f<-u;f+=i.length)_.set(i,f);i=_}if(c==="fusion")throw new Error(`Truncation strategy "${c}" not implemented`);d=await this._extract_fbank_features(i,this.mel_filters_slaney,this.config.nb_max_samples)}return d.unsqueeze_(0)}async _extract_fbank_features(i,l,c=null){return(0,o.spectrogram)(i,this.window,this.config.fft_window_size,this.config.hop_length,{power:2,mel_filters:l,log_mel:"dB",max_num_frames:c,do_pad:!1,transpose:!0})}async _call(i,{max_length:l=null}={}){return(0,s.validate_audio_inputs)(i,"ClapFeatureExtractor"),{input_features:(await this._get_input_mel(i,l??this.config.nb_max_samples,this.config.truncation,this.config.padding)).unsqueeze_(0)}}}},"./src/models/clip/image_processing_clip.js":(e,r,t)=>{t.r(r),t.d(r,{CLIPFeatureExtractor:()=>n,CLIPImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{}class n extends o{}},"./src/models/convnext/image_processing_convnext.js":(e,r,t)=>{t.r(r),t.d(r,{ConvNextFeatureExtractor:()=>n,ConvNextImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{constructor(i){super(i),this.crop_pct=this.config.crop_pct??224/256}async resize(i){var c;const l=(c=this.size)==null?void 0:c.shortest_edge;if(l===void 0)throw new Error("Size dictionary must contain 'shortest_edge' key.");if(l<384){const p=Math.floor(l/this.crop_pct),[d,u]=this.get_resize_output_image_size(i,{shortest_edge:p});i=await i.resize(d,u,{resample:this.resample}),i=await i.center_crop(l,l)}else i=await i.resize(l,l,{resample:this.resample});return i}}class n extends o{}},"./src/models/dac/feature_extraction_dac.js":(e,r,t)=>{t.r(r),t.d(r,{DacFeatureExtractor:()=>o});var s=t("./src/models/encodec/feature_extraction_encodec.js");class o extends s.EncodecFeatureExtractor{}},"./src/models/deit/image_processing_deit.js":(e,r,t)=>{t.r(r),t.d(r,{DeiTFeatureExtractor:()=>n,DeiTImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{}class n extends o{}},"./src/models/detr/image_processing_detr.js":(e,r,t)=>{t.r(r),t.d(r,{DetrFeatureExtractor:()=>a,DetrImageProcessor:()=>n});var s=t("./src/base/image_processors_utils.js"),o=t("./src/utils/tensor.js");class n extends s.ImageProcessor{async _call(l){const c=await super._call(l),p=[c.pixel_values.dims[0],64,64],d=(0,o.full)(p,1n);return{...c,pixel_mask:d}}post_process_object_detection(...l){return(0,s.post_process_object_detection)(...l)}post_process_panoptic_segmentation(...l){return(0,s.post_process_panoptic_segmentation)(...l)}post_process_instance_segmentation(...l){return(0,s.post_process_instance_segmentation)(...l)}}class a extends n{}},"./src/models/dinov3_vit/image_processing_dinov3_vit.js":(e,r,t)=>{t.r(r),t.d(r,{DINOv3ViTImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{}},"./src/models/donut/image_processing_donut.js":(e,r,t)=>{t.r(r),t.d(r,{DonutFeatureExtractor:()=>n,DonutImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{pad_image(i,l,c,p={}){const[d,u,_]=l;let f=this.image_mean;Array.isArray(this.image_mean)||(f=new Array(_).fill(f));let b=this.image_std;Array.isArray(b)||(b=new Array(_).fill(f));const A=f.map((g,y)=>-g/b[y]);return super.pad_image(i,l,c,{center:!0,constant_values:A,...p})}}class n extends o{}},"./src/models/dpt/image_processing_dpt.js":(e,r,t)=>{t.r(r),t.d(r,{DPTFeatureExtractor:()=>n,DPTImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{}class n extends o{}},"./src/models/efficientnet/image_processing_efficientnet.js":(e,r,t)=>{t.r(r),t.d(r,{EfficientNetImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{constructor(a){super(a),this.include_top=this.config.include_top??!0,this.include_top&&(this.image_std=this.image_std.map(i=>i*i))}}},"./src/models/encodec/feature_extraction_encodec.js":(e,r,t)=>{t.r(r),t.d(r,{EncodecFeatureExtractor:()=>n});var s=t("./src/base/feature_extraction_utils.js"),o=t("./src/utils/tensor.js");class n extends s.FeatureExtractor{async _call(i){(0,s.validate_audio_inputs)(i,"EncodecFeatureExtractor"),i instanceof Float64Array&&(i=new Float32Array(i));const l=this.config.feature_size;if(i.length%l!==0)throw new Error(`The length of the audio data must be a multiple of the number of channels (${l}).`);const c=[1,l,i.length/l];return{input_values:new o.Tensor("float32",i,c)}}}},"./src/models/feature_extractors.js":(e,r,t)=>{t.r(r),t.d(r,{ASTFeatureExtractor:()=>s.ASTFeatureExtractor,ClapFeatureExtractor:()=>n.ClapFeatureExtractor,DacFeatureExtractor:()=>a.DacFeatureExtractor,EncodecFeatureExtractor:()=>o.EncodecFeatureExtractor,Gemma3nAudioFeatureExtractor:()=>i.Gemma3nAudioFeatureExtractor,ImageFeatureExtractor:()=>g.ImageProcessor,MoonshineFeatureExtractor:()=>l.MoonshineFeatureExtractor,ParakeetFeatureExtractor:()=>c.ParakeetFeatureExtractor,PyAnnoteFeatureExtractor:()=>p.PyAnnoteFeatureExtractor,SeamlessM4TFeatureExtractor:()=>d.SeamlessM4TFeatureExtractor,SnacFeatureExtractor:()=>u.SnacFeatureExtractor,SpeechT5FeatureExtractor:()=>_.SpeechT5FeatureExtractor,Wav2Vec2FeatureExtractor:()=>f.Wav2Vec2FeatureExtractor,WeSpeakerFeatureExtractor:()=>b.WeSpeakerFeatureExtractor,WhisperFeatureExtractor:()=>A.WhisperFeatureExtractor});var s=t("./src/models/audio_spectrogram_transformer/feature_extraction_audio_spectrogram_transformer.js"),o=t("./src/models/encodec/feature_extraction_encodec.js"),n=t("./src/models/clap/feature_extraction_clap.js"),a=t("./src/models/dac/feature_extraction_dac.js"),i=t("./src/models/gemma3n/feature_extraction_gemma3n.js"),l=t("./src/models/moonshine/feature_extraction_moonshine.js"),c=t("./src/models/parakeet/feature_extraction_parakeet.js"),p=t("./src/models/pyannote/feature_extraction_pyannote.js"),d=t("./src/models/seamless_m4t/feature_extraction_seamless_m4t.js"),u=t("./src/models/snac/feature_extraction_snac.js"),_=t("./src/models/speecht5/feature_extraction_speecht5.js"),f=t("./src/models/wav2vec2/feature_extraction_wav2vec2.js"),b=t("./src/models/wespeaker/feature_extraction_wespeaker.js"),A=t("./src/models/whisper/feature_extraction_whisper.js"),g=t("./src/base/image_processors_utils.js")},"./src/models/florence2/processing_florence2.js":(e,r,t)=>{t.r(r),t.d(r,{Florence2Processor:()=>a});var s=t("./src/base/processing_utils.js"),o=t("./src/models/auto/image_processing_auto.js"),n=t("./src/tokenizers.js");class a extends s.Processor{constructor(l,c,p){super(l,c,p);const{tasks_answer_post_processing_type:d,task_prompts_without_inputs:u,task_prompts_with_input:_}=this.image_processor.config;this.tasks_answer_post_processing_type=new Map(Object.entries(d??{})),this.task_prompts_without_inputs=new Map(Object.entries(u??{})),this.task_prompts_with_input=new Map(Object.entries(_??{})),this.regexes={quad_boxes:/(.+?)<loc_(\d+)><loc_(\d+)><loc_(\d+)><loc_(\d+)><loc_(\d+)><loc_(\d+)><loc_(\d+)><loc_(\d+)>/gm,bboxes:/([^<]+)?<loc_(\d+)><loc_(\d+)><loc_(\d+)><loc_(\d+)>/gm},this.size_per_bin=1e3}construct_prompts(l){typeof l=="string"&&(l=[l]);const c=[];for(const p of l)if(this.task_prompts_without_inputs.has(p))c.push(this.task_prompts_without_inputs.get(p));else{for(const[d,u]of this.task_prompts_with_input)if(p.includes(d)){c.push(u.replaceAll("{input}",p).replaceAll(d,""));break}c.length!==l.length&&c.push(p)}return c}post_process_generation(l,c,p){const d=this.tasks_answer_post_processing_type.get(c)??"pure_text";l=l.replaceAll("<s>","").replaceAll("</s>","");let u;switch(d){case"pure_text":u=l;break;case"description_with_bboxes":case"bboxes":case"phrase_grounding":case"ocr":const _=d==="ocr"?"quad_boxes":"bboxes",f=l.matchAll(this.regexes[_]),b=[],A=[];for(const[g,y,...C]of f)b.push(y?y.trim():b.at(-1)??""),A.push(C.map((x,M)=>(Number(x)+.5)/this.size_per_bin*p[M%2]));u={labels:b,[_]:A};break;default:throw new Error(`Task "${c}" (of type "${d}") not yet implemented.`)}return{[c]:u}}async _call(l,c=null,p={}){if(!l&&!c)throw new Error("Either text or images must be provided");const d=await this.image_processor(l,p),u=c?this.tokenizer(this.construct_prompts(c),p):{};return{...d,...u}}}re(a,"tokenizer_class",n.AutoTokenizer),re(a,"image_processor_class",o.AutoImageProcessor)},"./src/models/gemma3n/feature_extraction_gemma3n.js":(e,r,t)=>{t.r(r),t.d(r,{Gemma3nAudioFeatureExtractor:()=>a});var s=t("./src/base/feature_extraction_utils.js"),o=t("./src/utils/tensor.js"),n=t("./src/utils/audio.js");class a extends s.FeatureExtractor{constructor(l){super(l);const{fft_length:c,feature_size:p,min_frequency:d,max_frequency:u,sampling_rate:_,frame_length:f}=this.config,b=(0,n.mel_filter_bank)(Math.floor(1+c/2),p,d,u,_,null,"htk",!1);this.mel_filters=b,this.window=(0,n.window_function)(f,"hann")}async _extract_fbank_features(l,c){return(0,n.spectrogram)(l,this.window,this.config.frame_length,this.config.hop_length,{fft_length:this.config.fft_length,center:!1,onesided:!0,preemphasis:this.config.preemphasis,preemphasis_htk_flavor:this.config.preemphasis_htk_flavor,mel_filters:this.mel_filters,log_mel:"log",mel_floor:this.config.mel_floor,remove_dc_offset:!1,transpose:!0})}async _call(l,{max_length:c=48e4,truncation:p=!0,padding:d=!0,pad_to_multiple_of:u=128}={}){if((0,s.validate_audio_inputs)(l,"Gemma3nAudioFeatureExtractor"),p&&l.length>c&&(l=l.slice(0,c)),d&&l.length%u!==0){const b=u-l.length%u,A=new Float64Array(l.length+b);A.set(l),this.config.padding_value!==0&&A.fill(this.config.padding_value,l.length),l=A}const _=await this._extract_fbank_features(l,this.config.max_length),f=(0,o.full)([1,_.dims[0]],!0);return{input_features:_.unsqueeze_(0),input_features_mask:f}}}},"./src/models/gemma3n/processing_gemma3n.js":(e,r,t)=>{t.r(r),t.d(r,{Gemma3nProcessor:()=>i});var s=t("./src/base/processing_utils.js"),o=t("./src/models/auto/image_processing_auto.js"),n=t("./src/models/auto/feature_extraction_auto.js"),a=t("./src/tokenizers.js");t("./src/utils/image.js"),t("./src/utils/audio.js");class i extends s.Processor{constructor(c,p,d){super(c,p,d),this.audio_seq_length=this.config.audio_seq_length,this.image_seq_length=this.config.image_seq_length;const{audio_token_id:u,boa_token:_,audio_token:f,eoa_token:b,image_token_id:A,boi_token:g,image_token:y,eoi_token:C}=this.tokenizer.config;this.audio_token_id=u,this.boa_token=_,this.audio_token=f;const x=f.repeat(this.audio_seq_length);this.full_audio_sequence=`

${_}${x}${b}

`,this.image_token_id=A,this.boi_token=g,this.image_token=y;const M=y.repeat(this.image_seq_length);this.full_image_sequence=`

${g}${M}${C}

`}async _call(c,p=null,d=null,u={}){typeof c=="string"&&(c=[c]);let _;d&&(_=await this.feature_extractor(d,u),c=c.map(A=>A.replaceAll(this.audio_token,this.full_audio_sequence)));let f;return p&&(f=await this.image_processor(p,u),c=c.map(A=>A.replaceAll(this.image_token,this.full_image_sequence))),{...this.tokenizer(c,u),...f,..._}}}re(i,"image_processor_class",o.AutoImageProcessor),re(i,"feature_extractor_class",n.AutoFeatureExtractor),re(i,"tokenizer_class",a.AutoTokenizer),re(i,"uses_processor_config",!0),re(i,"uses_chat_template_file",!0)},"./src/models/glpn/image_processing_glpn.js":(e,r,t)=>{t.r(r),t.d(r,{GLPNFeatureExtractor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{}},"./src/models/grounding_dino/image_processing_grounding_dino.js":(e,r,t)=>{t.r(r),t.d(r,{GroundingDinoImageProcessor:()=>n});var s=t("./src/base/image_processors_utils.js"),o=t("./src/utils/tensor.js");class n extends s.ImageProcessor{async _call(i){const l=await super._call(i),c=l.pixel_values.dims,p=(0,o.ones)([c[0],c[2],c[3]]);return{...l,pixel_mask:p}}}},"./src/models/grounding_dino/processing_grounding_dino.js":(e,r,t)=>{t.r(r),t.d(r,{GroundingDinoProcessor:()=>l});var s=t("./src/base/processing_utils.js"),o=t("./src/models/auto/image_processing_auto.js"),n=t("./src/tokenizers.js"),a=t("./src/base/image_processors_utils.js");function i(c,p){const u=c.dims.at(-1)-1,_=c.tolist();_.fill(!1,0,1),_.fill(!1,u);const f=p.tolist();return _.map((b,A)=>b?A:null).filter(b=>b!==null).map(b=>f[b])}class l extends s.Processor{async _call(p,d,u={}){const _=p?await this.image_processor(p,u):{};return{...d?this.tokenizer(d,u):{},..._}}post_process_grounded_object_detection(p,d,{box_threshold:u=.25,text_threshold:_=.25,target_sizes:f=null}={}){const{logits:b,pred_boxes:A}=p,g=b.dims[0];if(f!==null&&f.length!==g)throw Error("Make sure that you pass in as many target sizes as the batch dimension of the logits");const y=b.dims.at(1),C=b.sigmoid(),x=C.max(-1).tolist(),M=A.tolist().map(v=>v.map(P=>(0,a.center_to_corners_format)(P))),T=[];for(let v=0;v<g;++v){const P=f!==null?f[v]:null;P!==null&&(M[v]=M[v].map(j=>j.map((ne,q)=>ne*P[(q+1)%2])));const F=x[v],D=[],K=[],U=[];for(let j=0;j<y;++j){const ne=F[j];if(ne<=u)continue;const q=M[v][j],te=C[v][j];D.push(ne),U.push(q);const Z=i(te.gt(_),d[v]);K.push(Z)}T.push({scores:D,boxes:U,labels:this.batch_decode(K)})}return T}}re(l,"tokenizer_class",n.AutoTokenizer),re(l,"image_processor_class",o.AutoImageProcessor)},"./src/models/idefics3/image_processing_idefics3.js":(e,r,t)=>{t.r(r),t.d(r,{Idefics3ImageProcessor:()=>n});var s=t("./src/base/image_processors_utils.js"),o=t("./src/utils/tensor.js");class n extends s.ImageProcessor{constructor(i){super(i),this.do_image_splitting=i.do_image_splitting??!0,this.max_image_size=i.max_image_size}get_resize_for_vision_encoder(i,l){let[c,p]=i.dims.slice(-2);const d=p/c;return p>=c?(p=Math.ceil(p/l)*l,c=Math.floor(p/d),c=Math.ceil(c/l)*l):(c=Math.ceil(c/l)*l,p=Math.floor(c*d),p=Math.ceil(p/l)*l),{height:c,width:p}}async _call(i,{do_image_splitting:l=null,return_row_col_info:c=!1}={}){let p;if(!Array.isArray(i))p=[[i]];else{if(i.length===0||!i[0])throw new Error("No images provided.");Array.isArray(i[0])?p=i:p=[i]}let d=[],u=[],_=[];const f=[],b=[];for(const v of p){let P=await Promise.all(v.map(K=>this.preprocess(K)));f.push(...P.map(K=>K.original_size)),b.push(...P.map(K=>K.reshaped_input_size)),P.forEach(K=>K.pixel_values.unsqueeze_(0));const{longest_edge:F}=this.max_image_size;let D;if(l??this.do_image_splitting){let K=new Array(P.length),U=new Array(P.length);D=await Promise.all(P.map(async(j,ne)=>{const q=this.get_resize_for_vision_encoder(j.pixel_values,F),te=await(0,o.interpolate_4d)(j.pixel_values,{size:[q.height,q.width]}),{frames:Z,num_splits_h:ae,num_splits_w:he}=await this.split_image(te,this.max_image_size);return K[ne]=ae,U[ne]=he,(0,o.cat)(Z,0)})),u.push(K),_.push(U)}else{const K=[F,F];D=await Promise.all(P.map(U=>(0,o.interpolate_4d)(U.pixel_values,{size:K}))),u.push(new Array(P.length).fill(0)),_.push(new Array(P.length).fill(0))}d.push((0,o.cat)(D,0))}const A=d.length,[g,y,C,x]=d[0].dims;let M,T;if(A===1)M=d[0].unsqueeze_(0),T=(0,o.full)([A,g,C,x],!0);else{const v=Math.max(...d.map(D=>D.dims.at(0)));T=(0,o.full)([A,v,C,x],!0);const P=T.data,F=v*C*x;for(let D=0;D<A;++D){const K=d[D].dims[0];if(K<v){d[D]=(0,o.cat)([d[D],(0,o.full)([v-K,y,C,x],0)],0);const U=D*F+K*C*x,j=(D+1)*F;P.fill(!1,U,j)}}M=(0,o.stack)(d,0)}return{pixel_values:M,pixel_attention_mask:T,original_sizes:f,reshaped_input_sizes:b,...c?{rows:u,cols:_}:{}}}async split_image(i,{longest_edge:l}){const c=l,p=l,d=[],[u,_]=i.dims.slice(-2);let f=0,b=0;if(u>c||_>p){f=Math.ceil(u/c),b=Math.ceil(_/p);const A=Math.ceil(u/f),g=Math.ceil(_/b);for(let x=0;x<f;++x)for(let M=0;M<b;++M){let T,v,P,F;x===f-1?(v=u-A,F=u):(v=x*A,F=(x+1)*A),M===b-1?(T=_-g,P=_):(T=M*g,P=(M+1)*g);const D=[v,T],K=[F,P],U=await(0,o.slice)(i,D,K,[2,3]);d.push(U)}const y=c,C=p;(u!==y||_!==C)&&(i=await(0,o.interpolate_4d)(i,{size:[y,C]}))}return d.push(i),{frames:d,num_splits_h:f,num_splits_w:b}}}},"./src/models/idefics3/processing_idefics3.js":(e,r,t)=>{t.r(r),t.d(r,{Idefics3Processor:()=>p});var s=t("./src/base/processing_utils.js"),o=t("./src/models/auto/image_processing_auto.js"),n=t("./src/tokenizers.js");t("./src/utils/image.js");var a=t("./src/utils/core.js");function i(d,u,_,f,b,A){let g="";for(let y=0;y<u;++y){for(let C=0;C<_;++C)g+=f+`<row_${y+1}_col_${C+1}>`+b.repeat(d);g+=`
`}return g+=`
${f}${A}`+b.repeat(d)+`${f}`,g}function l(d,u,_,f){return`${u}${f}`+_.repeat(d)+`${u}`}function c(d,u,_,f,b,A){return d===0&&u===0?l(_,f,b,A):i(_,d,u,f,b,A)}class p extends s.Processor{constructor(){super(...arguments);re(this,"fake_image_token","<fake_token_around_image>");re(this,"image_token","<image>");re(this,"global_img_token","<global-img>")}async _call(_,f=null,b={}){b.return_row_col_info??(b.return_row_col_info=!0);let A;f&&(A=await this.image_processor(f,b)),Array.isArray(_)||(_=[_]);const g=A.rows??[new Array(_.length).fill(0)],y=A.cols??[new Array(_.length).fill(0)],C=this.config.image_seq_len,x=[],M=[];for(let v=0;v<_.length;++v){const P=_[v],F=g[v],D=y[v];x.push((0,a.count)(P,this.image_token));const K=F.map((ne,q)=>c(ne,D[q],C,this.fake_image_token,this.image_token,this.global_img_token)),U=P.split(this.image_token);if(U.length===0)throw new Error("The image token should be present in the text.");let j=U[0];for(let ne=0;ne<K.length;++ne)j+=K[ne]+U[ne+1];M.push(j)}return{...this.tokenizer(M),...A}}}re(p,"image_processor_class",o.AutoImageProcessor),re(p,"tokenizer_class",n.AutoTokenizer),re(p,"uses_processor_config",!0)},"./src/models/image_processors.js":(e,r,t)=>{t.r(r),t.d(r,{BeitFeatureExtractor:()=>s.BeitFeatureExtractor,BitImageProcessor:()=>o.BitImageProcessor,CLIPFeatureExtractor:()=>a.CLIPFeatureExtractor,CLIPImageProcessor:()=>a.CLIPImageProcessor,ChineseCLIPFeatureExtractor:()=>n.ChineseCLIPFeatureExtractor,ConvNextFeatureExtractor:()=>i.ConvNextFeatureExtractor,ConvNextImageProcessor:()=>i.ConvNextImageProcessor,DINOv3ViTImageProcessor:()=>p.DINOv3ViTImageProcessor,DPTFeatureExtractor:()=>u.DPTFeatureExtractor,DPTImageProcessor:()=>u.DPTImageProcessor,DeiTFeatureExtractor:()=>l.DeiTFeatureExtractor,DeiTImageProcessor:()=>l.DeiTImageProcessor,DetrFeatureExtractor:()=>c.DetrFeatureExtractor,DetrImageProcessor:()=>c.DetrImageProcessor,DonutFeatureExtractor:()=>d.DonutFeatureExtractor,DonutImageProcessor:()=>d.DonutImageProcessor,EfficientNetImageProcessor:()=>_.EfficientNetImageProcessor,GLPNFeatureExtractor:()=>f.GLPNFeatureExtractor,GroundingDinoImageProcessor:()=>b.GroundingDinoImageProcessor,Idefics3ImageProcessor:()=>A.Idefics3ImageProcessor,JinaCLIPImageProcessor:()=>y.JinaCLIPImageProcessor,LlavaOnevisionImageProcessor:()=>C.LlavaOnevisionImageProcessor,Mask2FormerImageProcessor:()=>x.Mask2FormerImageProcessor,MaskFormerFeatureExtractor:()=>M.MaskFormerFeatureExtractor,MaskFormerImageProcessor:()=>M.MaskFormerImageProcessor,MobileNetV1FeatureExtractor:()=>T.MobileNetV1FeatureExtractor,MobileNetV1ImageProcessor:()=>T.MobileNetV1ImageProcessor,MobileNetV2FeatureExtractor:()=>v.MobileNetV2FeatureExtractor,MobileNetV2ImageProcessor:()=>v.MobileNetV2ImageProcessor,MobileNetV3FeatureExtractor:()=>P.MobileNetV3FeatureExtractor,MobileNetV3ImageProcessor:()=>P.MobileNetV3ImageProcessor,MobileNetV4FeatureExtractor:()=>F.MobileNetV4FeatureExtractor,MobileNetV4ImageProcessor:()=>F.MobileNetV4ImageProcessor,MobileViTFeatureExtractor:()=>D.MobileViTFeatureExtractor,MobileViTImageProcessor:()=>D.MobileViTImageProcessor,NougatImageProcessor:()=>K.NougatImageProcessor,OwlViTFeatureExtractor:()=>j.OwlViTFeatureExtractor,OwlViTImageProcessor:()=>j.OwlViTImageProcessor,Owlv2ImageProcessor:()=>U.Owlv2ImageProcessor,Phi3VImageProcessor:()=>ne.Phi3VImageProcessor,PvtImageProcessor:()=>q.PvtImageProcessor,Qwen2VLImageProcessor:()=>te.Qwen2VLImageProcessor,RTDetrImageProcessor:()=>Z.RTDetrImageProcessor,SamImageProcessor:()=>ae.SamImageProcessor,SegformerFeatureExtractor:()=>he.SegformerFeatureExtractor,SegformerImageProcessor:()=>he.SegformerImageProcessor,SiglipImageProcessor:()=>Q.SiglipImageProcessor,SmolVLMImageProcessor:()=>B.SmolVLMImageProcessor,Swin2SRImageProcessor:()=>O.Swin2SRImageProcessor,VLMImageProcessor:()=>g.VLMImageProcessor,ViTFeatureExtractor:()=>W.ViTFeatureExtractor,ViTImageProcessor:()=>W.ViTImageProcessor,VitMatteImageProcessor:()=>N.VitMatteImageProcessor,VitPoseImageProcessor:()=>J.VitPoseImageProcessor,YolosFeatureExtractor:()=>ie.YolosFeatureExtractor,YolosImageProcessor:()=>ie.YolosImageProcessor});var s=t("./src/models/beit/image_processing_beit.js"),o=t("./src/models/bit/image_processing_bit.js"),n=t("./src/models/chinese_clip/image_processing_chinese_clip.js"),a=t("./src/models/clip/image_processing_clip.js"),i=t("./src/models/convnext/image_processing_convnext.js"),l=t("./src/models/deit/image_processing_deit.js"),c=t("./src/models/detr/image_processing_detr.js"),p=t("./src/models/dinov3_vit/image_processing_dinov3_vit.js"),d=t("./src/models/donut/image_processing_donut.js"),u=t("./src/models/dpt/image_processing_dpt.js"),_=t("./src/models/efficientnet/image_processing_efficientnet.js"),f=t("./src/models/glpn/image_processing_glpn.js"),b=t("./src/models/grounding_dino/image_processing_grounding_dino.js"),A=t("./src/models/idefics3/image_processing_idefics3.js"),g=t("./src/models/janus/image_processing_janus.js"),y=t("./src/models/jina_clip/image_processing_jina_clip.js"),C=t("./src/models/llava_onevision/image_processing_llava_onevision.js"),x=t("./src/models/mask2former/image_processing_mask2former.js"),M=t("./src/models/maskformer/image_processing_maskformer.js"),T=t("./src/models/mobilenet_v1/image_processing_mobilenet_v1.js"),v=t("./src/models/mobilenet_v2/image_processing_mobilenet_v2.js"),P=t("./src/models/mobilenet_v3/image_processing_mobilenet_v3.js"),F=t("./src/models/mobilenet_v4/image_processing_mobilenet_v4.js"),D=t("./src/models/mobilevit/image_processing_mobilevit.js"),K=t("./src/models/nougat/image_processing_nougat.js"),U=t("./src/models/owlv2/image_processing_owlv2.js"),j=t("./src/models/owlvit/image_processing_owlvit.js"),ne=t("./src/models/phi3_v/image_processing_phi3_v.js"),q=t("./src/models/pvt/image_processing_pvt.js"),te=t("./src/models/qwen2_vl/image_processing_qwen2_vl.js"),Z=t("./src/models/rt_detr/image_processing_rt_detr.js"),ae=t("./src/models/sam/image_processing_sam.js"),he=t("./src/models/segformer/image_processing_segformer.js"),Q=t("./src/models/siglip/image_processing_siglip.js"),B=t("./src/models/smolvlm/image_processing_smolvlm.js"),O=t("./src/models/swin2sr/image_processing_swin2sr.js"),W=t("./src/models/vit/image_processing_vit.js"),N=t("./src/models/vitmatte/image_processing_vitmatte.js"),J=t("./src/models/vitpose/image_processing_vitpose.js"),ie=t("./src/models/yolos/image_processing_yolos.js")},"./src/models/janus/image_processing_janus.js":(e,r,t)=>{t.r(r),t.d(r,{VLMImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{constructor(a){super({do_pad:!0,pad_size:{width:a.image_size,height:a.image_size},...a}),this.constant_values=this.config.background_color.map(i=>i*this.rescale_factor)}pad_image(a,i,l,c){return super.pad_image(a,i,l,{constant_values:this.constant_values,center:!0,...c})}}},"./src/models/janus/processing_janus.js":(e,r,t)=>{t.r(r),t.d(r,{VLChatProcessor:()=>c});var s=t("./src/base/processing_utils.js"),o=t("./src/models/auto/image_processing_auto.js"),n=t("./src/tokenizers.js"),a=t("./src/utils/core.js"),i=t("./src/utils/tensor.js"),l=t("./src/utils/image.js");class c extends s.Processor{constructor(d,u,_){super(d,u,_),this.image_tag=this.config.image_tag,this.image_start_tag=this.config.image_start_tag,this.image_end_tag=this.config.image_end_tag,this.num_image_tokens=this.config.num_image_tokens}async _call(d,{images:u=null,chat_template:_="default"}={}){u?Array.isArray(u)||(u=[u]):u=await Promise.all(d.filter(D=>D.images).flatMap(D=>D.images).map(D=>l.RawImage.read(D)));const f=this.tokenizer,b=f.apply_chat_template(d,{tokenize:!1,add_generation_prompt:!0,chat_template:_}),A=D=>f.encode(D,{add_special_tokens:!1}),g=b.split(this.image_tag),y=g.length-1;if(u.length!==y)throw new Error(`Number of images provided (${u.length}) does not match number of "${this.image_tag}" image tags (${y})`);const[C,x,M]=f.model.convert_tokens_to_ids([this.image_tag,this.image_start_tag,this.image_end_tag]);let T=A(g[0]),v=new Array(T.length).fill(!1);for(let D=1;D<g.length;++D){const K=new Array(this.num_image_tokens).fill(C),U=A(g[D]);T=(0,a.mergeArrays)(T,[x],K,[M],U);const j=new Array(this.num_image_tokens).fill(!0);v=(0,a.mergeArrays)(v,[!1],j,[!1],new Array(U.length).fill(!1))}const P=[1,T.length],F={input_ids:new i.Tensor("int64",T,P),attention_mask:new i.Tensor("int64",new Array(T.length).fill(1),P),images_seq_mask:new i.Tensor("bool",v,P),images_emb_mask:new i.Tensor("bool",new Array(y*this.num_image_tokens).fill(!0),[1,y,this.num_image_tokens])};if(u&&u.length>0){const D=await this.image_processor(u);return D.pixel_values.unsqueeze_(0),{...F,...D}}return F}}re(c,"image_processor_class",o.AutoImageProcessor),re(c,"tokenizer_class",n.AutoTokenizer),re(c,"uses_processor_config",!0)},"./src/models/jina_clip/image_processing_jina_clip.js":(e,r,t)=>{t.r(r),t.d(r,{JinaCLIPImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{constructor(a){const{resize_mode:i,fill_color:l,interpolation:c,size:p,...d}=a,u=i==="squash"?{width:p,height:p}:i==="shortest"?{shortest_edge:p}:{longest_edge:p},_=c==="bicubic"?3:2;super({...d,size:u,resample:_,do_center_crop:!0,crop_size:p,do_normalize:!0})}}},"./src/models/jina_clip/processing_jina_clip.js":(e,r,t)=>{t.r(r),t.d(r,{JinaCLIPProcessor:()=>a});var s=t("./src/base/processing_utils.js"),o=t("./src/models/auto/image_processing_auto.js"),n=t("./src/tokenizers.js");class a extends s.Processor{async _call(l=null,c=null,p={}){if(!l&&!c)throw new Error("Either text or images must be provided");const d=l?this.tokenizer(l,p):{},u=c?await this.image_processor(c,p):{};return{...d,...u}}}re(a,"tokenizer_class",n.AutoTokenizer),re(a,"image_processor_class",o.AutoImageProcessor)},"./src/models/llava/processing_llava.js":(e,r,t)=>{t.r(r),t.d(r,{LlavaProcessor:()=>a});var s=t("./src/base/processing_utils.js"),o=t("./src/models/auto/image_processing_auto.js"),n=t("./src/tokenizers.js");class a extends s.Processor{async _call(l,c=null,p={}){const d=await this.image_processor(l,p);if(c){const[_,f]=d.pixel_values.dims.slice(-2),{image_token:b,patch_size:A,num_additional_image_tokens:g}=this.config,y=Math.floor(_/A)*Math.floor(f/A)+g;c=structuredClone(c),Array.isArray(c)||(c=[c]);for(let C=0;C<c.length;++C)c[C]=c[C].replace(b,b.repeat(y))}const u=c?this.tokenizer(c,p):{};return{...d,...u}}}re(a,"tokenizer_class",n.AutoTokenizer),re(a,"image_processor_class",o.AutoImageProcessor),re(a,"uses_processor_config",!0)},"./src/models/llava_onevision/image_processing_llava_onevision.js":(e,r,t)=>{t.r(r),t.d(r,{LlavaOnevisionImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{}},"./src/models/mask2former/image_processing_mask2former.js":(e,r,t)=>{t.r(r),t.d(r,{Mask2FormerImageProcessor:()=>o});var s=t("./src/models/maskformer/image_processing_maskformer.js");class o extends s.MaskFormerImageProcessor{}},"./src/models/maskformer/image_processing_maskformer.js":(e,r,t)=>{t.r(r),t.d(r,{MaskFormerFeatureExtractor:()=>n,MaskFormerImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{post_process_panoptic_segmentation(...i){return(0,s.post_process_panoptic_segmentation)(...i)}post_process_instance_segmentation(...i){return(0,s.post_process_instance_segmentation)(...i)}}class n extends o{}},"./src/models/mgp_str/processing_mgp_str.js":(e,r,t)=>{t.r(r),t.d(r,{MgpstrProcessor:()=>l});var s=t("./src/base/processing_utils.js"),o=t("./src/models/auto/image_processing_auto.js"),n=t("./src/tokenizers.js"),a=t("./src/utils/maths.js");const i={char:["char_decode",1],bpe:["bpe_decode",2],wp:["wp_decode",102]};class l extends s.Processor{get char_tokenizer(){return this.components.char_tokenizer}get bpe_tokenizer(){return this.components.bpe_tokenizer}get wp_tokenizer(){return this.components.wp_tokenizer}_decode_helper(p,d){if(!i.hasOwnProperty(d))throw new Error(`Format ${d} is not supported.`);const[u,_]=i[d],f=this[u].bind(this),[b,A]=p.dims,g=[],y=[],C=p.tolist();for(let M=0;M<b;++M){const T=C[M],v=[],P=[];for(let D=1;D<A;++D){const[K,U]=(0,a.max)((0,a.softmax)(T[D]));if(P.push(K),U==_)break;v.push(U)}const F=P.length>0?P.reduce((D,K)=>D*K,1):0;y.push(v),g.push(F)}return[f(y),g]}char_decode(p){return this.char_tokenizer.batch_decode(p).map(d=>d.replaceAll(" ",""))}bpe_decode(p){return this.bpe_tokenizer.batch_decode(p)}wp_decode(p){return this.wp_tokenizer.batch_decode(p).map(d=>d.replaceAll(" ",""))}batch_decode([p,d,u]){const[_,f]=this._decode_helper(p,"char"),[b,A]=this._decode_helper(d,"bpe"),[g,y]=this._decode_helper(u,"wp"),C=[],x=[];for(let M=0;M<_.length;++M){const[T,v]=(0,a.max)([f[M],A[M],y[M]]);C.push([_[M],b[M],g[M]][v]),x.push(T)}return{generated_text:C,scores:x,char_preds:_,bpe_preds:b,wp_preds:g}}static async from_pretrained(...p){const d=await super.from_pretrained(...p),u=await n.AutoTokenizer.from_pretrained("Xenova/gpt2"),_=await n.AutoTokenizer.from_pretrained("Xenova/bert-base-uncased");return d.components={image_processor:d.image_processor,char_tokenizer:d.tokenizer,bpe_tokenizer:u,wp_tokenizer:_},d}async _call(p,d=null){const u=await this.image_processor(p);return d&&(u.labels=this.tokenizer(d).input_ids),u}}re(l,"tokenizer_class",n.AutoTokenizer),re(l,"image_processor_class",o.AutoImageProcessor)},"./src/models/mobilenet_v1/image_processing_mobilenet_v1.js":(e,r,t)=>{t.r(r),t.d(r,{MobileNetV1FeatureExtractor:()=>n,MobileNetV1ImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{}class n extends o{}},"./src/models/mobilenet_v2/image_processing_mobilenet_v2.js":(e,r,t)=>{t.r(r),t.d(r,{MobileNetV2FeatureExtractor:()=>n,MobileNetV2ImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{}class n extends o{}},"./src/models/mobilenet_v3/image_processing_mobilenet_v3.js":(e,r,t)=>{t.r(r),t.d(r,{MobileNetV3FeatureExtractor:()=>n,MobileNetV3ImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{}class n extends o{}},"./src/models/mobilenet_v4/image_processing_mobilenet_v4.js":(e,r,t)=>{t.r(r),t.d(r,{MobileNetV4FeatureExtractor:()=>n,MobileNetV4ImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{}class n extends o{}},"./src/models/mobilevit/image_processing_mobilevit.js":(e,r,t)=>{t.r(r),t.d(r,{MobileViTFeatureExtractor:()=>n,MobileViTImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{}class n extends o{}},"./src/models/moonshine/feature_extraction_moonshine.js":(e,r,t)=>{t.r(r),t.d(r,{MoonshineFeatureExtractor:()=>n});var s=t("./src/base/feature_extraction_utils.js"),o=t("./src/utils/tensor.js");class n extends s.FeatureExtractor{async _call(i){(0,s.validate_audio_inputs)(i,"MoonshineFeatureExtractor"),i instanceof Float64Array&&(i=new Float32Array(i));const l=[1,i.length];return{input_values:new o.Tensor("float32",i,l)}}}},"./src/models/moonshine/processing_moonshine.js":(e,r,t)=>{t.r(r),t.d(r,{MoonshineProcessor:()=>a});var s=t("./src/models/auto/feature_extraction_auto.js"),o=t("./src/tokenizers.js"),n=t("./src/base/processing_utils.js");class a extends n.Processor{async _call(l){return await this.feature_extractor(l)}}re(a,"tokenizer_class",o.AutoTokenizer),re(a,"feature_extractor_class",s.AutoFeatureExtractor)},"./src/models/nougat/image_processing_nougat.js":(e,r,t)=>{t.r(r),t.d(r,{NougatImageProcessor:()=>o});var s=t("./src/models/donut/image_processing_donut.js");class o extends s.DonutImageProcessor{}},"./src/models/owlv2/image_processing_owlv2.js":(e,r,t)=>{t.r(r),t.d(r,{Owlv2ImageProcessor:()=>o});var s=t("./src/models/owlvit/image_processing_owlvit.js");class o extends s.OwlViTImageProcessor{}},"./src/models/owlvit/image_processing_owlvit.js":(e,r,t)=>{t.r(r),t.d(r,{OwlViTFeatureExtractor:()=>n,OwlViTImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{post_process_object_detection(...i){return(0,s.post_process_object_detection)(...i)}}class n extends o{}},"./src/models/owlvit/processing_owlvit.js":(e,r,t)=>{t.r(r),t.d(r,{OwlViTProcessor:()=>a});var s=t("./src/base/processing_utils.js"),o=t("./src/models/auto/image_processing_auto.js"),n=t("./src/tokenizers.js");class a extends s.Processor{}re(a,"tokenizer_class",n.AutoTokenizer),re(a,"image_processor_class",o.AutoImageProcessor)},"./src/models/paligemma/processing_paligemma.js":(e,r,t)=>{t.r(r),t.d(r,{PaliGemmaProcessor:()=>l});var s=t("./src/base/processing_utils.js"),o=t("./src/models/auto/image_processing_auto.js"),n=t("./src/tokenizers.js");const a="<image>";function i(c,p,d,u,_){return`${u.repeat(d*_)}${p}${c}
`}class l extends s.Processor{async _call(p,d=null,u={}){d||(console.warn("You are using PaliGemma without a text prefix. It will perform as a picture-captioning model."),d=""),Array.isArray(p)||(p=[p]),Array.isArray(d)||(d=[d]);const _=this.tokenizer.bos_token,f=this.image_processor.config.image_seq_length;let b;d.some(y=>y.includes(a))?b=d.map(y=>{const C=y.replaceAll(a,a.repeat(f)),x=C.lastIndexOf(a),M=x===-1?0:x+a.length;return C.slice(0,M)+_+C.slice(M)+`
`}):(console.warn("You are passing both `text` and `images` to `PaliGemmaProcessor`. The processor expects special image tokens in the text, as many tokens as there are images per each text. It is recommended to add `<image>` tokens in the very beginning of your text. For this call, we will infer how many images each text has and add special tokens."),b=d.map(y=>i(y,_,f,a,p.length)));const A=this.tokenizer(b,u);return{...await this.image_processor(p,u),...A}}}re(l,"tokenizer_class",n.AutoTokenizer),re(l,"image_processor_class",o.AutoImageProcessor),re(l,"uses_processor_config",!1)},"./src/models/parakeet/feature_extraction_parakeet.js":(e,r,t)=>{t.r(r),t.d(r,{ParakeetFeatureExtractor:()=>i});var s=t("./src/base/feature_extraction_utils.js"),o=t("./src/utils/tensor.js"),n=t("./src/utils/audio.js");const a=1e-5;class i extends s.FeatureExtractor{constructor(c){var u;super(c),(u=this.config).mel_filters??(u.mel_filters=(0,n.mel_filter_bank)(Math.floor(1+this.config.n_fft/2),this.config.feature_size,0,this.config.sampling_rate/2,this.config.sampling_rate,"slaney","slaney"));const p=(0,n.window_function)(this.config.win_length,"hann",{periodic:!1});this.window=new Float64Array(this.config.n_fft);const d=Math.floor((this.config.n_fft-this.config.win_length)/2);this.window.set(p,d)}async _extract_fbank_features(c){const p=this.config.preemphasis;c=new Float64Array(c);for(let u=c.length-1;u>=1;--u)c[u]-=p*c[u-1];return await(0,n.spectrogram)(c,this.window,this.window.length,this.config.hop_length,{fft_length:this.config.n_fft,power:2,mel_filters:this.config.mel_filters,log_mel:"log",mel_floor:-1/0,pad_mode:"constant",center:!0,transpose:!0,mel_offset:2**-24})}async _call(c){(0,s.validate_audio_inputs)(c,"ParakeetFeatureExtractor");const p=await this._extract_fbank_features(c),d=Math.floor((c.length+Math.floor(this.config.n_fft/2)*2-this.config.n_fft)/this.config.hop_length),u=p.data;u.fill(0,d*p.dims[1]);const[_,f]=p.dims,b=new Float64Array(f),A=new Float64Array(f);for(let C=0;C<d;++C){const x=C*f;for(let M=0;M<f;++M){const T=u[x+M];b[M]+=T,A[M]+=T*T}}const g=d>1?d-1:1;for(let C=0;C<f;++C){const x=b[C]/d,M=(A[C]-d*x*x)/g,v=1/(Math.sqrt(M)+a);for(let P=0;P<d;++P){const F=P*f+C;u[F]=(u[F]-x)*v}}const y=new BigInt64Array(_);return y.fill(1n,0,d),{input_features:p.unsqueeze_(0),attention_mask:new o.Tensor("int64",y,[1,_])}}}},"./src/models/phi3_v/image_processing_phi3_v.js":(e,r,t)=>{t.r(r),t.d(r,{Phi3VImageProcessor:()=>p});var s=t("./src/base/image_processors_utils.js"),o=t("./src/utils/tensor.js");const n=336,a=[2,3],{ceil:i,floor:l,sqrt:c}=Math;class p extends s.ImageProcessor{constructor(u){super({...u,do_normalize:!0,do_pad:!0,pad_size:"custom",do_convert_rgb:!0,do_resize:!0}),this._num_crops=u.num_crops}calc_num_image_tokens_from_image_size(u,_){const{num_img_tokens:f}=this.config;return l((l(_/n)*l(u/n)+1)*f+1+(l(_/n)+1)*c(f))}get_resize_output_image_size(u,_){const f=this._num_crops,[b,A]=u.size;let g=b/A,y=1;for(;y*Math.ceil(y/g)<=f;)y+=1;y-=1;const C=Math.floor(y*336),x=Math.floor(C/g);return[C,x]}pad_image(u,_,f,b={}){const[A,g]=_,y=n*i(A/n),C=n*i(g/n),x=[1,1,1].map((M,T)=>(M-this.image_mean[T])/this.image_std[T]);return super.pad_image(u,_,{width:C,height:y},{center:!0,constant_values:x,...b})}async _call(u,{num_crops:_=null}={}){if(this._num_crops=_??(_=this.config.num_crops),_<4||c(_)%1!==0)throw new Error("num_crops must be a square number >= 4");Array.isArray(u)||(u=[u]);const f=u.length,b=await Promise.all(u.map(v=>this.preprocess(v))),A=b.map(v=>v.original_size),g=b.map(v=>v.reshaped_input_size),y=[];for(const{pixel_values:v}of b){v.unsqueeze_(0);const[P,F]=v.dims.slice(-2),D=await(0,o.interpolate_4d)(v,{size:[n,n],mode:"bicubic"});if(_>0){const K=[],U=c(_),j=l(F/U),ne=l(P/U);for(let te=0;te<U;++te)for(let Z=0;Z<U;++Z){let ae,he,Q,B;te===U-1?(he=P-ne,B=P):(he=te*ne,B=(te+1)*ne),Z===U-1?(ae=F-j,Q=F):(ae=Z*j,Q=(Z+1)*j);const O=[he,ae],W=[B,Q],N=await(0,o.slice)(v,O,W,a);K.push(N)}const q=await(0,o.interpolate_4d)((0,o.cat)(K,0),{size:[n,n],mode:"bicubic"});y.push((0,o.cat)([D,q],0))}else y.push(D)}const C=(0,o.stack)(y,0),x=g.map(v=>v.map(P=>n*i(P/n))),M=new o.Tensor("int64",x.flat(),[f,2]),T=x.map(([v,P])=>this.calc_num_image_tokens_from_image_size(P,v));return{pixel_values:C,original_sizes:A,reshaped_input_sizes:g,image_sizes:M,num_img_tokens:T}}}},"./src/models/phi3_v/processing_phi3_v.js":(e,r,t)=>{t.r(r),t.d(r,{Phi3VProcessor:()=>l});var s=t("./src/base/processing_utils.js"),o=t("./src/models/auto/image_processing_auto.js"),n=t("./src/tokenizers.js");t("./src/utils/image.js");const a="<|image|>",i=/<\|image_\d+\|>/g;class l extends s.Processor{async _call(p,d=null,{padding:u=!0,truncation:_=!0,num_crops:f=null}={}){Array.isArray(p)||(p=[p]);let b,A;if(d){A=await this.image_processor(d,{num_crops:f});const{num_img_tokens:g}=A,y=p.map((x,M)=>x.split(i).join(a.repeat(g[M])));b=this.tokenizer(y,{padding:u,truncation:_});const C=this.tokenizer.model.convert_tokens_to_ids([a])[0];b.input_ids.map_(x=>x==C?-x:x)}else b=this.tokenizer(p);return{...b,...A}}}re(l,"image_processor_class",o.AutoImageProcessor),re(l,"tokenizer_class",n.AutoTokenizer)},"./src/models/processors.js":(e,r,t)=>{t.r(r),t.d(r,{Florence2Processor:()=>s.Florence2Processor,Gemma3nProcessor:()=>o.Gemma3nProcessor,GroundingDinoProcessor:()=>n.GroundingDinoProcessor,Idefics3Processor:()=>a.Idefics3Processor,JinaCLIPProcessor:()=>l.JinaCLIPProcessor,LlavaProcessor:()=>c.LlavaProcessor,MgpstrProcessor:()=>p.MgpstrProcessor,MoonshineProcessor:()=>d.MoonshineProcessor,OwlViTProcessor:()=>u.OwlViTProcessor,PaliGemmaProcessor:()=>f.PaliGemmaProcessor,Phi3VProcessor:()=>_.Phi3VProcessor,PyAnnoteProcessor:()=>b.PyAnnoteProcessor,Qwen2VLProcessor:()=>A.Qwen2VLProcessor,SamProcessor:()=>g.SamProcessor,SmolVLMProcessor:()=>y.SmolVLMProcessor,SpeechT5Processor:()=>C.SpeechT5Processor,UltravoxProcessor:()=>x.UltravoxProcessor,VLChatProcessor:()=>i.VLChatProcessor,VoxtralProcessor:()=>M.VoxtralProcessor,Wav2Vec2Processor:()=>T.Wav2Vec2Processor,Wav2Vec2ProcessorWithLM:()=>v.Wav2Vec2ProcessorWithLM,WhisperProcessor:()=>P.WhisperProcessor});var s=t("./src/models/florence2/processing_florence2.js"),o=t("./src/models/gemma3n/processing_gemma3n.js"),n=t("./src/models/grounding_dino/processing_grounding_dino.js"),a=t("./src/models/idefics3/processing_idefics3.js"),i=t("./src/models/janus/processing_janus.js"),l=t("./src/models/jina_clip/processing_jina_clip.js"),c=t("./src/models/llava/processing_llava.js"),p=t("./src/models/mgp_str/processing_mgp_str.js"),d=t("./src/models/moonshine/processing_moonshine.js"),u=t("./src/models/owlvit/processing_owlvit.js"),_=t("./src/models/phi3_v/processing_phi3_v.js"),f=t("./src/models/paligemma/processing_paligemma.js"),b=t("./src/models/pyannote/processing_pyannote.js"),A=t("./src/models/qwen2_vl/processing_qwen2_vl.js"),g=t("./src/models/sam/processing_sam.js"),y=t("./src/models/smolvlm/processing_smolvlm.js"),C=t("./src/models/speecht5/processing_speecht5.js"),x=t("./src/models/ultravox/processing_ultravox.js"),M=t("./src/models/voxtral/processing_voxtral.js"),T=t("./src/models/wav2vec2/processing_wav2vec2.js"),v=t("./src/models/wav2vec2_with_lm/processing_wav2vec2_with_lm.js"),P=t("./src/models/whisper/processing_whisper.js")},"./src/models/pvt/image_processing_pvt.js":(e,r,t)=>{t.r(r),t.d(r,{PvtImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{}},"./src/models/pyannote/feature_extraction_pyannote.js":(e,r,t)=>{t.r(r),t.d(r,{PyAnnoteFeatureExtractor:()=>a});var s=t("./src/base/feature_extraction_utils.js"),o=t("./src/utils/tensor.js"),n=t("./src/utils/maths.js");class a extends s.FeatureExtractor{async _call(l){(0,s.validate_audio_inputs)(l,"PyAnnoteFeatureExtractor"),l instanceof Float64Array&&(l=new Float32Array(l));const c=[1,1,l.length];return{input_values:new o.Tensor("float32",l,c)}}samples_to_frames(l){return(l-this.config.offset)/this.config.step}post_process_speaker_diarization(l,c){const p=c/this.samples_to_frames(c)/this.config.sampling_rate,d=[];for(const u of l.tolist()){const _=[];let f=-1;for(let b=0;b<u.length;++b){const A=(0,n.softmax)(u[b]),[g,y]=(0,n.max)(A),[C,x]=[b,b+1];y!==f?(f=y,_.push({id:y,start:C,end:x,score:g})):(_.at(-1).end=x,_.at(-1).score+=g)}d.push(_.map(({id:b,start:A,end:g,score:y})=>({id:b,start:A*p,end:g*p,confidence:y/(g-A)})))}return d}}},"./src/models/pyannote/processing_pyannote.js":(e,r,t)=>{t.r(r),t.d(r,{PyAnnoteProcessor:()=>n});var s=t("./src/base/processing_utils.js"),o=t("./src/models/pyannote/feature_extraction_pyannote.js");class n extends s.Processor{async _call(i){return await this.feature_extractor(i)}post_process_speaker_diarization(...i){return this.feature_extractor.post_process_speaker_diarization(...i)}get sampling_rate(){return this.feature_extractor.config.sampling_rate}}re(n,"feature_extractor_class",o.PyAnnoteFeatureExtractor)},"./src/models/qwen2_vl/image_processing_qwen2_vl.js":(e,r,t)=>{t.r(r),t.d(r,{Qwen2VLImageProcessor:()=>n});var s=t("./src/base/image_processors_utils.js"),o=t("./src/utils/tensor.js");class n extends s.ImageProcessor{async _call(i,...l){const{pixel_values:c,original_sizes:p,reshaped_input_sizes:d}=await super._call(i,...l);let u=c;const{temporal_patch_size:_,merge_size:f,patch_size:b}=this.config;u.dims[0]===1&&(u=(0,o.cat)(Array.from({length:_},()=>u),0));const A=u.dims[0]/_,g=u.dims[1],y=Math.floor(u.dims[2]/b),C=Math.floor(u.dims[3]/b),x=u.view(A,_,g,Math.floor(y/f),f,b,Math.floor(C/f),f,b).permute(0,3,6,4,7,2,1,5,8).view(A*y*C,g*_*b*b),M=new o.Tensor("int64",[A,y,C],[1,3]);return{pixel_values:x,image_grid_thw:M,original_sizes:p,reshaped_input_sizes:d}}}},"./src/models/qwen2_vl/processing_qwen2_vl.js":(e,r,t)=>{t.r(r),t.d(r,{Qwen2VLProcessor:()=>a});var s=t("./src/base/processing_utils.js"),o=t("./src/models/auto/image_processing_auto.js"),n=t("./src/tokenizers.js");t("./src/utils/image.js");class a extends s.Processor{async _call(l,c=null,...p){Array.isArray(l)||(l=[l]);let d,u;if(c&&(d=await this.image_processor(c),u=d.image_grid_thw),u){let f=this.image_processor.config.merge_size**2,b=0;const A=u.tolist();l=l.map(g=>{for(;g.includes("<|image_pad|>");){const y=Number(A[b++].reduce((C,x)=>C*x,1n));g=g.replace("<|image_pad|>","<|placeholder|>".repeat(Math.floor(y/f)))}return g.replaceAll("<|placeholder|>","<|image_pad|>")})}return{...this.tokenizer(l),...d}}}re(a,"image_processor_class",o.AutoImageProcessor),re(a,"tokenizer_class",n.AutoTokenizer)},"./src/models/rt_detr/image_processing_rt_detr.js":(e,r,t)=>{t.r(r),t.d(r,{RTDetrImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{post_process_object_detection(...a){return(0,s.post_process_object_detection)(...a)}}},"./src/models/sam/image_processing_sam.js":(e,r,t)=>{t.r(r),t.d(r,{SamImageProcessor:()=>a});var s=t("./src/base/image_processors_utils.js"),o=t("./src/utils/core.js"),n=t("./src/utils/tensor.js");class a extends s.ImageProcessor{reshape_input_points(l,c,p,d=!1){l=structuredClone(l);let u=(0,o.calculateDimensions)(l);if(u.length===3)d||(u=[1,...u]),l=[l];else if(u.length!==4)throw Error("The input_points must be a 4D tensor of shape `batch_size`, `point_batch_size`, `nb_points_per_image`, `2`.");for(let _=0;_<l.length;++_){let f=c[_],b=p[_],A=[b[0]/f[0],b[1]/f[1]];for(let g=0;g<l[_].length;++g)for(let y=0;y<l[_][g].length;++y)for(let C=0;C<l[_][g][y].length;++C)l[_][g][y][C]*=A[C%2]}return new n.Tensor("float32",Float32Array.from(l.flat(1/0)),u)}add_input_labels(l,c){let p=(0,o.calculateDimensions)(l);if(p.length===2)p=[1,...p],l=[l];else if(p.length!==3)throw Error("The input_points must be a 4D tensor of shape `batch_size`, `point_batch_size`, `nb_points_per_image`, `2`.");if(p.some((d,u)=>d!==c.dims[u]))throw Error(`The first ${p.length} dimensions of 'input_points' and 'input_labels' must be the same.`);return new n.Tensor("int64",l.flat(1/0).map(BigInt),p)}async _call(l,{input_points:c=null,input_labels:p=null,input_boxes:d=null}={}){const u=await super._call(l);if(c&&(u.input_points=this.reshape_input_points(c,u.original_sizes,u.reshaped_input_sizes)),p){if(!u.input_points)throw Error("`input_points` must be provided if `input_labels` are provided.");u.input_labels=this.add_input_labels(p,u.input_points)}return d&&(u.input_boxes=this.reshape_input_points(d,u.original_sizes,u.reshaped_input_sizes,!0)),u}async post_process_masks(l,c,p,{mask_threshold:d=0,binarize:u=!0,pad_size:_=null}={}){const f=[];_=_??this.pad_size;const b=[_.height,_.width];for(let A=0;A<c.length;++A){const g=c[A],y=p[A];let C=await(0,n.interpolate_4d)(l[A],{mode:"bilinear",size:b});if(C=C.slice(null,null,[0,y[0]],[0,y[1]]),C=await(0,n.interpolate_4d)(C,{mode:"bilinear",size:g}),u){const x=C.data,M=new Uint8Array(x.length);for(let T=0;T<x.length;++T)x[T]>d&&(M[T]=1);C=new n.Tensor("bool",M,C.dims)}f.push(C)}return f}generate_crop_boxes(l,c,{crop_n_layers:p=0,overlap_ratio:d=512/1500,points_per_crop:u=32,crop_n_points_downscale_factor:_=1}={}){}}},"./src/models/sam/processing_sam.js":(e,r,t)=>{t.r(r),t.d(r,{SamProcessor:()=>n});var s=t("./src/base/processing_utils.js"),o=t("./src/models/auto/image_processing_auto.js");class n extends s.Processor{async _call(...i){return await this.image_processor(...i)}post_process_masks(...i){return this.image_processor.post_process_masks(...i)}reshape_input_points(...i){return this.image_processor.reshape_input_points(...i)}}re(n,"image_processor_class",o.AutoImageProcessor)},"./src/models/seamless_m4t/feature_extraction_seamless_m4t.js":(e,r,t)=>{t.r(r),t.d(r,{SeamlessM4TFeatureExtractor:()=>a});var s=t("./src/base/feature_extraction_utils.js"),o=t("./src/utils/tensor.js"),n=t("./src/utils/audio.js");class a extends s.FeatureExtractor{constructor(l){super(l);const c=this.config.sampling_rate,p=(0,n.mel_filter_bank)(257,this.config.num_mel_bins,20,Math.floor(c/2),c,null,"kaldi",!0);this.mel_filters=p,this.window=(0,n.window_function)(400,"povey",{periodic:!1})}async _extract_fbank_features(l,c){return l=l.map(p=>p*32768),(0,n.spectrogram)(l,this.window,400,160,{fft_length:512,power:2,center:!1,preemphasis:.97,mel_filters:this.mel_filters,log_mel:"log",mel_floor:1192092955078125e-22,remove_dc_offset:!0,max_num_frames:c,transpose:!0})}async _call(l,{padding:c=!0,pad_to_multiple_of:p=2,do_normalize_per_mel_bins:d=!0,return_attention_mask:u=!0}={}){(0,s.validate_audio_inputs)(l,"SeamlessM4TFeatureExtractor");let _=await this._extract_fbank_features(l,this.config.max_length);if(d){const[M,T]=_.dims,v=_.data;for(let P=0;P<T;++P){let F=0;for(let j=0;j<M;++j)F+=v[j*T+P];const D=F/M;let K=0;for(let j=0;j<M;++j)K+=(v[j*T+P]-D)**2;K/=M-1;const U=Math.sqrt(K+1e-7);for(let j=0;j<M;++j){const ne=j*T+P;v[ne]=(v[ne]-D)/U}}}let f;if(c){const[M,T]=_.dims,v=_.data,P=M%p;if(P>0){const F=new Float32Array(T*(M+P));F.set(v),F.fill(this.config.padding_value,v.length);const D=M+P;_=new o.Tensor(_.type,F,[D,T]),u&&(f=new o.Tensor("int64",new BigInt64Array(D),[1,D]),f.data.fill(1n,0,M))}}const[b,A]=_.dims,g=this.config.stride;if(b%g!==0)throw new Error(`The number of frames (${b}) must be a multiple of the stride (${g}).`);const C=_.view(1,Math.floor(b/g),A*g),x={input_features:C};if(u){const M=C.dims[1],T=new BigInt64Array(M);if(f){const v=f.data;for(let P=1,F=0;P<b;P+=g,++F)T[F]=v[P]}else T.fill(1n);x.attention_mask=new o.Tensor("int64",T,[1,M])}return x}}},"./src/models/segformer/image_processing_segformer.js":(e,r,t)=>{t.r(r),t.d(r,{SegformerFeatureExtractor:()=>n,SegformerImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{post_process_semantic_segmentation(...i){return(0,s.post_process_semantic_segmentation)(...i)}}class n extends o{}},"./src/models/siglip/image_processing_siglip.js":(e,r,t)=>{t.r(r),t.d(r,{SiglipImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{}},"./src/models/smolvlm/image_processing_smolvlm.js":(e,r,t)=>{t.r(r),t.d(r,{SmolVLMImageProcessor:()=>s.Idefics3ImageProcessor});var s=t("./src/models/idefics3/image_processing_idefics3.js")},"./src/models/smolvlm/processing_smolvlm.js":(e,r,t)=>{t.r(r),t.d(r,{SmolVLMProcessor:()=>s.Idefics3Processor});var s=t("./src/models/idefics3/processing_idefics3.js")},"./src/models/snac/feature_extraction_snac.js":(e,r,t)=>{t.r(r),t.d(r,{SnacFeatureExtractor:()=>o});var s=t("./src/models/dac/feature_extraction_dac.js");class o extends s.DacFeatureExtractor{}},"./src/models/speecht5/feature_extraction_speecht5.js":(e,r,t)=>{t.r(r),t.d(r,{SpeechT5FeatureExtractor:()=>o});var s=t("./src/base/feature_extraction_utils.js");class o extends s.FeatureExtractor{}},"./src/models/speecht5/processing_speecht5.js":(e,r,t)=>{t.r(r),t.d(r,{SpeechT5Processor:()=>a});var s=t("./src/base/processing_utils.js"),o=t("./src/tokenizers.js"),n=t("./src/models/auto/feature_extraction_auto.js");class a extends s.Processor{async _call(l){return await this.feature_extractor(l)}}re(a,"tokenizer_class",o.AutoTokenizer),re(a,"feature_extractor_class",n.AutoFeatureExtractor)},"./src/models/swin2sr/image_processing_swin2sr.js":(e,r,t)=>{t.r(r),t.d(r,{Swin2SRImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{pad_image(a,i,l,c={}){const[p,d,u]=i;return super.pad_image(a,i,{width:d+(l-d%l)%l,height:p+(l-p%l)%l},{mode:"symmetric",center:!1,constant_values:-1,...c})}}},"./src/models/ultravox/processing_ultravox.js":(e,r,t)=>{t.r(r),t.d(r,{UltravoxProcessor:()=>a});var s=t("./src/models/auto/feature_extraction_auto.js"),o=t("./src/tokenizers.js"),n=t("./src/base/processing_utils.js");class a extends n.Processor{async _call(l,c=null,p={}){if(Array.isArray(l))throw new Error("Batched inputs are not supported yet.");let d={};if(c){const _=c.length,{input_features:f}=await this.feature_extractor(c,{...p,max_length:_}),b=Math.round(_/this.config.encoder_ds_factor+1e-4),A=1+Math.ceil(b/this.config.stack_factor);d.audio_token_len=[A],d.audio_values=f;const g=this.config.audio_placeholder;if(!l.includes(g))throw new Error(`The input text does not contain the image token ${g}.`);l=l.replaceAll(g,g.repeat(A))}return{...this.tokenizer(l,{add_special_tokens:!1,...p}),...d}}}re(a,"tokenizer_class",o.AutoTokenizer),re(a,"feature_extractor_class",s.AutoFeatureExtractor),re(a,"uses_processor_config",!0)},"./src/models/vit/image_processing_vit.js":(e,r,t)=>{t.r(r),t.d(r,{ViTFeatureExtractor:()=>n,ViTImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{}class n extends o{}},"./src/models/vitmatte/image_processing_vitmatte.js":(e,r,t)=>{t.r(r),t.d(r,{VitMatteImageProcessor:()=>n});var s=t("./src/base/image_processors_utils.js"),o=t("./src/utils/tensor.js");class n extends s.ImageProcessor{async _call(i,l){Array.isArray(i)||(i=[i]),Array.isArray(l)||(l=[l]);const c=await Promise.all(i.map(u=>this.preprocess(u))),p=await Promise.all(l.map(u=>this.preprocess(u,{do_normalize:!1,do_convert_rgb:!1,do_convert_grayscale:!0})));return{pixel_values:(0,o.stack)(c.map((u,_)=>(0,o.cat)([u.pixel_values,p[_].pixel_values],0)),0),original_sizes:c.map(u=>u.original_size),reshaped_input_sizes:c.map(u=>u.reshaped_input_size)}}}},"./src/models/vitpose/image_processing_vitpose.js":(e,r,t)=>{t.r(r),t.d(r,{VitPoseImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{post_process_pose_estimation(a,i,{threshold:l=null}={}){const c=a.tolist(),[p,d,u,_]=a.dims,f=[];for(let b=0;b<p;++b){const A=c[b],g=i[b],y=[];for(let C=0;C<g.length;++C){const x=g[C],M=[],T=[],v=[],P=x.at(-2)/_,F=x.at(-1)/u;for(let D=0;D<A.length;++D){let[K,U]=[0,0],j=0,ne=-1/0;const q=A[D];for(let Z=0;Z<q.length;++Z){const ae=q[Z];for(let he=0;he<ae.length;++he){const Q=ae[he];j+=Q,ne=Math.max(ne,Q),K+=(he+.5)*Q,U+=Z*Q}}if(l!=null&&ne<l)continue;const te=[P*K/j,F*U/j];M.push(te),v.push(D),T.push(ne)}y.push({bbox:x,scores:T,labels:v,keypoints:M})}f.push(y)}return f}}},"./src/models/voxtral/processing_voxtral.js":(e,r,t)=>{t.r(r),t.d(r,{VoxtralProcessor:()=>d});var s=t("./src/models/auto/feature_extraction_auto.js"),o=t("./src/tokenizers.js"),n=t("./src/base/processing_utils.js"),a=t("./src/utils/tensor.js");const i="[AUDIO]",l="[BEGIN_AUDIO]",c=375;function p(u,_){const f=[];for(let b=0;b<u.length;b+=_)f.push(u.subarray(b,Math.min(b+_,u.length)));return f}class d extends n.Processor{async _call(_,f=null,b={}){if(Array.isArray(_))throw new Error("Batched inputs are not supported yet.");const A={};if(f){if(!_.includes(i))throw new Error(`The input text does not contain the audio token ${i}.`);Array.isArray(f)||(f=[f]);const y=_.split(i),C=y.length-1;if(C!==f.length)throw new Error(`The number of audio inputs (${f.length}) does not match the number of audio tokens in the text (${C}).`);const x=this.feature_extractor.config.n_samples,M=f.map(D=>p(D,x)),T=M.map(D=>D.length),v=M.flat(),P=(await Promise.all(v.map(D=>this.feature_extractor(D,b)))).map(D=>D.input_features);A.audio_values=P.length>1?(0,a.cat)(P,0):P[0];let F=y[0];for(let D=0;D<T.length;++D){F+=l;for(let K=0;K<T[D];++K)F+=i.repeat(c);F+=y[D+1]}_=F}return{...this.tokenizer(_,{add_special_tokens:!1,...b}),...A}}}re(d,"tokenizer_class",o.AutoTokenizer),re(d,"feature_extractor_class",s.AutoFeatureExtractor),re(d,"uses_processor_config",!1)},"./src/models/wav2vec2/feature_extraction_wav2vec2.js":(e,r,t)=>{t.r(r),t.d(r,{Wav2Vec2FeatureExtractor:()=>n});var s=t("./src/base/feature_extraction_utils.js"),o=t("./src/utils/tensor.js");class n extends s.FeatureExtractor{_zero_mean_unit_var_norm(i){const c=i.reduce((d,u)=>d+u,0)/i.length,p=i.reduce((d,u)=>d+(u-c)**2,0)/i.length;return i.map(d=>(d-c)/Math.sqrt(p+1e-7))}async _call(i){(0,s.validate_audio_inputs)(i,"Wav2Vec2FeatureExtractor"),i instanceof Float64Array&&(i=new Float32Array(i));let l=i;this.config.do_normalize&&(l=this._zero_mean_unit_var_norm(l));const c=[1,l.length];return{input_values:new o.Tensor("float32",l,c),attention_mask:new o.Tensor("int64",new BigInt64Array(l.length).fill(1n),c)}}}},"./src/models/wav2vec2/processing_wav2vec2.js":(e,r,t)=>{t.r(r),t.d(r,{Wav2Vec2Processor:()=>a});var s=t("./src/tokenizers.js"),o=t("./src/models/auto/feature_extraction_auto.js"),n=t("./src/base/processing_utils.js");class a extends n.Processor{async _call(l){return await this.feature_extractor(l)}}re(a,"tokenizer_class",s.AutoTokenizer),re(a,"feature_extractor_class",o.AutoFeatureExtractor)},"./src/models/wav2vec2_with_lm/processing_wav2vec2_with_lm.js":(e,r,t)=>{t.r(r),t.d(r,{Wav2Vec2ProcessorWithLM:()=>a});var s=t("./src/tokenizers.js"),o=t("./src/models/auto/feature_extraction_auto.js"),n=t("./src/base/processing_utils.js");class a extends n.Processor{async _call(l){return await this.feature_extractor(l)}}re(a,"tokenizer_class",s.AutoTokenizer),re(a,"feature_extractor_class",o.AutoFeatureExtractor)},"./src/models/wespeaker/feature_extraction_wespeaker.js":(e,r,t)=>{t.r(r),t.d(r,{WeSpeakerFeatureExtractor:()=>n});var s=t("./src/base/feature_extraction_utils.js");t("./src/utils/tensor.js");var o=t("./src/utils/audio.js");class n extends s.FeatureExtractor{constructor(i){super(i);const l=this.config.sampling_rate,c=(0,o.mel_filter_bank)(257,this.config.num_mel_bins,20,Math.floor(l/2),l,null,"kaldi",!0);this.mel_filters=c,this.window=(0,o.window_function)(400,"hamming",{periodic:!1}),this.min_num_frames=this.config.min_num_frames}async _extract_fbank_features(i){return i=i.map(l=>l*32768),(0,o.spectrogram)(i,this.window,400,160,{fft_length:512,power:2,center:!1,preemphasis:.97,mel_filters:this.mel_filters,log_mel:"log",mel_floor:1192092955078125e-22,remove_dc_offset:!0,transpose:!0,min_num_frames:this.min_num_frames})}async _call(i){(0,s.validate_audio_inputs)(i,"WeSpeakerFeatureExtractor");const l=(await this._extract_fbank_features(i)).unsqueeze_(0);if(this.config.fbank_centering_span===null){const c=l.mean(1).data,p=l.data,[d,u,_]=l.dims;for(let f=0;f<d;++f){const b=f*u*_,A=f*_;for(let g=0;g<u;++g){const y=b+g*_;for(let C=0;C<_;++C)p[y+C]-=c[A+C]}}}return{input_features:l}}}},"./src/models/whisper/common_whisper.js":(e,r,t)=>{t.r(r),t.d(r,{WHISPER_LANGUAGE_MAPPING:()=>o,WHISPER_TO_LANGUAGE_CODE_MAPPING:()=>n,whisper_language_to_code:()=>a});const s=[["en","english"],["zh","chinese"],["de","german"],["es","spanish"],["ru","russian"],["ko","korean"],["fr","french"],["ja","japanese"],["pt","portuguese"],["tr","turkish"],["pl","polish"],["ca","catalan"],["nl","dutch"],["ar","arabic"],["sv","swedish"],["it","italian"],["id","indonesian"],["hi","hindi"],["fi","finnish"],["vi","vietnamese"],["he","hebrew"],["uk","ukrainian"],["el","greek"],["ms","malay"],["cs","czech"],["ro","romanian"],["da","danish"],["hu","hungarian"],["ta","tamil"],["no","norwegian"],["th","thai"],["ur","urdu"],["hr","croatian"],["bg","bulgarian"],["lt","lithuanian"],["la","latin"],["mi","maori"],["ml","malayalam"],["cy","welsh"],["sk","slovak"],["te","telugu"],["fa","persian"],["lv","latvian"],["bn","bengali"],["sr","serbian"],["az","azerbaijani"],["sl","slovenian"],["kn","kannada"],["et","estonian"],["mk","macedonian"],["br","breton"],["eu","basque"],["is","icelandic"],["hy","armenian"],["ne","nepali"],["mn","mongolian"],["bs","bosnian"],["kk","kazakh"],["sq","albanian"],["sw","swahili"],["gl","galician"],["mr","marathi"],["pa","punjabi"],["si","sinhala"],["km","khmer"],["sn","shona"],["yo","yoruba"],["so","somali"],["af","afrikaans"],["oc","occitan"],["ka","georgian"],["be","belarusian"],["tg","tajik"],["sd","sindhi"],["gu","gujarati"],["am","amharic"],["yi","yiddish"],["lo","lao"],["uz","uzbek"],["fo","faroese"],["ht","haitian creole"],["ps","pashto"],["tk","turkmen"],["nn","nynorsk"],["mt","maltese"],["sa","sanskrit"],["lb","luxembourgish"],["my","myanmar"],["bo","tibetan"],["tl","tagalog"],["mg","malagasy"],["as","assamese"],["tt","tatar"],["haw","hawaiian"],["ln","lingala"],["ha","hausa"],["ba","bashkir"],["jw","javanese"],["su","sundanese"]],o=new Map(s),n=new Map([...s.map(([i,l])=>[l,i]),["burmese","my"],["valencian","ca"],["flemish","nl"],["haitian","ht"],["letzeburgesch","lb"],["pushto","ps"],["panjabi","pa"],["moldavian","ro"],["moldovan","ro"],["sinhalese","si"],["castilian","es"]]);function a(i){i=i.toLowerCase();let l=n.get(i);if(l===void 0){const c=i.match(/^<\|([a-z]{2})\|>$/);if(c&&(i=c[1]),o.has(i))l=i;else{const d=i.length===2?o.keys():o.values();throw new Error(`Language "${i}" is not supported. Must be one of: ${JSON.stringify(Array.from(d))}`)}}return l}},"./src/models/whisper/feature_extraction_whisper.js":(e,r,t)=>{t.r(r),t.d(r,{WhisperFeatureExtractor:()=>a});var s=t("./src/base/feature_extraction_utils.js");t("./src/utils/tensor.js");var o=t("./src/utils/audio.js"),n=t("./src/utils/maths.js");class a extends s.FeatureExtractor{constructor(l){var c;super(l),(c=this.config).mel_filters??(c.mel_filters=(0,o.mel_filter_bank)(Math.floor(1+this.config.n_fft/2),this.config.feature_size,0,8e3,this.config.sampling_rate,"slaney","slaney")),this.window=(0,o.window_function)(this.config.n_fft,"hann")}async _extract_fbank_features(l){const c=await(0,o.spectrogram)(l,this.window,this.config.n_fft,this.config.hop_length,{power:2,mel_filters:this.config.mel_filters,log_mel:"log10",max_num_frames:Math.min(Math.floor(l.length/this.config.hop_length),this.config.nb_max_frames)}),p=c.data,d=(0,n.max)(p)[0];for(let u=0;u<p.length;++u)p[u]=(Math.max(p[u],d-8)+4)/4;return c}async _call(l,{max_length:c=null}={}){(0,s.validate_audio_inputs)(l,"WhisperFeatureExtractor");let p;const d=c??this.config.n_samples;return l.length>d?(l.length>this.config.n_samples&&console.warn("Attempting to extract features for audio longer than 30 seconds. If using a pipeline to extract transcript from a long audio clip, remember to specify `chunk_length_s` and/or `stride_length_s`."),p=l.slice(0,d)):(p=new Float32Array(d),p.set(l)),{input_features:(await this._extract_fbank_features(p)).unsqueeze_(0)}}}},"./src/models/whisper/generation_whisper.js":(e,r,t)=>{t.r(r),t.d(r,{WhisperGenerationConfig:()=>o});var s=t("./src/generation/configuration_utils.js");class o extends s.GenerationConfig{constructor(){super(...arguments);re(this,"return_timestamps",null);re(this,"return_token_timestamps",null);re(this,"num_frames",null);re(this,"alignment_heads",null);re(this,"task",null);re(this,"language",null);re(this,"no_timestamps_token_id",null);re(this,"prompt_ids",null);re(this,"is_multilingual",null);re(this,"lang_to_id",null);re(this,"task_to_id",null);re(this,"max_initial_timestamp_index",1)}}},"./src/models/whisper/processing_whisper.js":(e,r,t)=>{t.r(r),t.d(r,{WhisperProcessor:()=>a});var s=t("./src/models/auto/feature_extraction_auto.js"),o=t("./src/tokenizers.js"),n=t("./src/base/processing_utils.js");class a extends n.Processor{async _call(l){return await this.feature_extractor(l)}}re(a,"tokenizer_class",o.AutoTokenizer),re(a,"feature_extractor_class",s.AutoFeatureExtractor)},"./src/models/yolos/image_processing_yolos.js":(e,r,t)=>{t.r(r),t.d(r,{YolosFeatureExtractor:()=>n,YolosImageProcessor:()=>o});var s=t("./src/base/image_processors_utils.js");class o extends s.ImageProcessor{post_process_object_detection(...i){return(0,s.post_process_object_detection)(...i)}}class n extends o{}},"./src/ops/registry.js":(e,r,t)=>{t.r(r),t.d(r,{TensorOpRegistry:()=>a});var s=t("./src/backends/onnx.js"),o=t("./src/utils/tensor.js");const n=async(i,l,c)=>{const p=await(0,s.createInferenceSession)(new Uint8Array(i),l);return async d=>{const u=(0,s.isONNXProxy)(),_=Object.fromEntries(Object.entries(d).map(([b,A])=>[b,(u?A.clone():A).ort_tensor])),f=await(0,s.runInferenceSession)(p,_);return Array.isArray(c)?c.map(b=>new o.Tensor(f[b])):new o.Tensor(f[c])}};class a{static get nearest_interpolate_4d(){return this._nearest_interpolate_4d||(this._nearest_interpolate_4d=n([8,10,18,0,58,129,1,10,41,10,1,120,10,0,10,0,10,1,115,18,1,121,34,6,82,101,115,105,122,101,42,18,10,4,109,111,100,101,34,7,110,101,97,114,101,115,116,160,1,3,18,1,114,90,31,10,1,120,18,26,10,24,8,1,18,20,10,3,18,1,98,10,3,18,1,99,10,3,18,1,104,10,3,18,1,119,90,15,10,1,115,18,10,10,8,8,7,18,4,10,2,8,4,98,31,10,1,121,18,26,10,24,8,1,18,20,10,3,18,1,98,10,3,18,1,99,10,3,18,1,104,10,3,18,1,119,66,2,16,21],this.session_options,"y")),this._nearest_interpolate_4d}static get bilinear_interpolate_4d(){return this._bilinear_interpolate_4d||(this._bilinear_interpolate_4d=n([8,9,18,0,58,128,1,10,40,10,1,120,10,0,10,0,10,1,115,18,1,121,34,6,82,101,115,105,122,101,42,17,10,4,109,111,100,101,34,6,108,105,110,101,97,114,160,1,3,18,1,114,90,31,10,1,120,18,26,10,24,8,1,18,20,10,3,18,1,98,10,3,18,1,99,10,3,18,1,104,10,3,18,1,119,90,15,10,1,115,18,10,10,8,8,7,18,4,10,2,8,4,98,31,10,1,121,18,26,10,24,8,1,18,20,10,3,18,1,98,10,3,18,1,99,10,3,18,1,104,10,3,18,1,119,66,2,16,20],this.session_options,"y")),this._bilinear_interpolate_4d}static get bicubic_interpolate_4d(){return this._bicubic_interpolate_4d||(this._bicubic_interpolate_4d=n([8,9,18,0,58,127,10,39,10,1,120,10,0,10,0,10,1,115,18,1,121,34,6,82,101,115,105,122,101,42,16,10,4,109,111,100,101,34,5,99,117,98,105,99,160,1,3,18,1,114,90,31,10,1,120,18,26,10,24,8,1,18,20,10,3,18,1,98,10,3,18,1,99,10,3,18,1,104,10,3,18,1,119,90,15,10,1,115,18,10,10,8,8,7,18,4,10,2,8,4,98,31,10,1,121,18,26,10,24,8,1,18,20,10,3,18,1,98,10,3,18,1,99,10,3,18,1,104,10,3,18,1,119,66,2,16,20],this.session_options,"y")),this._bicubic_interpolate_4d}static get matmul(){return this._matmul||(this._matmul=n([8,9,18,0,58,55,10,17,10,1,97,10,1,98,18,1,99,34,6,77,97,116,77,117,108,18,1,114,90,9,10,1,97,18,4,10,2,8,1,90,9,10,1,98,18,4,10,2,8,1,98,9,10,1,99,18,4,10,2,8,1,66,2,16,20],this.session_options,"c")),this._matmul}static get stft(){return this._stft||(this._stft=n([8,7,18,0,58,148,1,10,38,10,1,115,10,1,106,10,1,119,10,1,108,18,1,111,34,4,83,84,70,84,42,15,10,8,111,110,101,115,105,100,101,100,24,1,160,1,2,18,1,115,90,26,10,1,115,18,21,10,19,8,1,18,15,10,3,18,1,98,10,3,18,1,115,10,3,18,1,99,90,11,10,1,106,18,6,10,4,8,7,18,0,90,16,10,1,119,18,11,10,9,8,1,18,5,10,3,18,1,119,90,11,10,1,108,18,6,10,4,8,7,18,0,98,31,10,1,111,18,26,10,24,8,1,18,20,10,3,18,1,98,10,3,18,1,102,10,3,18,1,100,10,3,18,1,99,66,2,16,17],this.session_options,"o")),this._stft}static get rfft(){return this._rfft||(this._rfft=n([8,9,18,0,58,97,10,33,10,1,120,10,0,10,1,97,18,1,121,34,3,68,70,84,42,15,10,8,111,110,101,115,105,100,101,100,24,1,160,1,2,18,1,100,90,21,10,1,120,18,16,10,14,8,1,18,10,10,3,18,1,115,10,3,18,1,99,90,11,10,1,97,18,6,10,4,8,7,18,0,98,21,10,1,121,18,16,10,14,8,1,18,10,10,3,18,1,115,10,3,18,1,99,66,2,16,20],this.session_options,"y")),this._rfft}static get top_k(){return this._top_k||(this._top_k=n([8,10,18,0,58,73,10,18,10,1,120,10,1,107,18,1,118,18,1,105,34,4,84,111,112,75,18,1,116,90,9,10,1,120,18,4,10,2,8,1,90,15,10,1,107,18,10,10,8,8,7,18,4,10,2,8,1,98,9,10,1,118,18,4,10,2,8,1,98,9,10,1,105,18,4,10,2,8,7,66,2,16,21],this.session_options,["v","i"])),this._top_k}static get slice(){return this._slice||(this._slice=n([8,7,18,0,58,96,10,25,10,1,120,10,1,115,10,1,101,10,1,97,10,1,116,18,1,121,34,5,83,108,105,99,101,18,1,114,90,9,10,1,120,18,4,10,2,8,1,90,9,10,1,115,18,4,10,2,8,7,90,9,10,1,101,18,4,10,2,8,7,90,9,10,1,97,18,4,10,2,8,7,90,9,10,1,116,18,4,10,2,8,7,98,9,10,1,121,18,4,10,2,8,1,66,2,16,13],this.session_options,"y")),this._slice}}re(a,"session_options",{})},"./src/pipelines.js":(e,r,t)=>{t.r(r),t.d(r,{AudioClassificationPipeline:()=>U,AutomaticSpeechRecognitionPipeline:()=>ne,BackgroundRemovalPipeline:()=>ae,DepthEstimationPipeline:()=>J,DocumentQuestionAnsweringPipeline:()=>O,FeatureExtractionPipeline:()=>D,FillMaskPipeline:()=>C,ImageClassificationPipeline:()=>te,ImageFeatureExtractionPipeline:()=>K,ImageSegmentationPipeline:()=>Z,ImageToImagePipeline:()=>N,ImageToTextPipeline:()=>q,ObjectDetectionPipeline:()=>Q,Pipeline:()=>b,QuestionAnsweringPipeline:()=>y,SummarizationPipeline:()=>M,Text2TextGenerationPipeline:()=>x,TextClassificationPipeline:()=>A,TextGenerationPipeline:()=>P,TextToAudioPipeline:()=>W,TokenClassificationPipeline:()=>g,TranslationPipeline:()=>T,ZeroShotAudioClassificationPipeline:()=>j,ZeroShotClassificationPipeline:()=>F,ZeroShotImageClassificationPipeline:()=>he,ZeroShotObjectDetectionPipeline:()=>B,pipeline:()=>Ae});var s=t("./src/tokenizers.js"),o=t("./src/models.js"),n=t("./src/models/auto/processing_auto.js");t("./src/base/processing_utils.js");var a=t("./src/utils/generic.js"),i=t("./src/utils/core.js"),l=t("./src/utils/maths.js"),c=t("./src/utils/audio.js"),p=t("./src/utils/tensor.js"),d=t("./src/utils/image.js");async function u($e){return Array.isArray($e)||($e=[$e]),await Promise.all($e.map(X=>d.RawImage.read(X)))}async function _($e,X){return Array.isArray($e)||($e=[$e]),await Promise.all($e.map(z=>typeof z=="string"||z instanceof URL?(0,c.read_audio)(z,X):z instanceof Float64Array?new Float32Array(z):z))}function f($e,X){X&&($e=$e.map(Ce=>Ce|0));const[z,_e,Ee,Me]=$e;return{xmin:z,ymin:_e,xmax:Ee,ymax:Me}}class b extends a.Callable{constructor({task:X,model:z,tokenizer:_e=null,processor:Ee=null}){super(),this.task=X,this.model=z,this.tokenizer=_e,this.processor=Ee}async dispose(){await this.model.dispose()}}class A extends b{constructor(X){super(X)}async _call(X,{top_k:z=1}={}){const _e=this.tokenizer(X,{padding:!0,truncation:!0}),Ee=await this.model(_e),Me=this.model.config.problem_type==="multi_label_classification"?de=>de.sigmoid():de=>new p.Tensor("float32",(0,l.softmax)(de.data),de.dims),Ce=this.model.config.id2label,ye=[];for(const de of Ee.logits){const we=Me(de),ce=await(0,p.topk)(we,z),ke=ce[0].tolist(),Te=ce[1].tolist().map((We,qe)=>({label:Ce?Ce[We]:`LABEL_${We}`,score:ke[qe]}));z===1?ye.push(...Te):ye.push(Te)}return Array.isArray(X)||z===1?ye:ye[0]}}class g extends b{constructor(X){super(X)}async _call(X,{ignore_labels:z=["O"]}={}){const _e=Array.isArray(X),Ee=this.tokenizer(_e?X:[X],{padding:!0,truncation:!0}),Ce=(await this.model(Ee)).logits,ye=this.model.config.id2label,de=[];for(let we=0;we<Ce.dims[0];++we){const ce=Ee.input_ids[we],ke=Ce[we],Le=[];for(let Te=0;Te<ke.dims[0];++Te){const We=ke[Te],qe=(0,l.max)(We.data)[1],st=ye?ye[qe]:`LABEL_${qe}`;if(z.includes(st))continue;const Ze=this.tokenizer.decode([ce[Te].item()],{skip_special_tokens:!0});if(Ze==="")continue;const ze=(0,l.softmax)(We.data);Le.push({entity:st,score:ze[qe],index:Te,word:Ze})}de.push(Le)}return _e?de:de[0]}}class y extends b{constructor(X){super(X)}async _call(X,z,{top_k:_e=1}={}){const Ee=this.tokenizer(X,{text_pair:z,padding:!0,truncation:!0}),{start_logits:Me,end_logits:Ce}=await this.model(Ee),ye=Ee.input_ids.tolist(),de=Ee.attention_mask.tolist(),we=this.tokenizer.all_special_ids,ce=[];for(let ke=0;ke<Me.dims[0];++ke){const Le=ye[ke],Te=Le.findIndex(He=>He==this.tokenizer.sep_token_id);de[ke].map((He,gt)=>He==1&&(gt===0||gt>Te&&we.findIndex(dt=>dt==Le[gt])===-1));const We=Me[ke].tolist(),qe=Ce[ke].tolist();for(let He=1;He<We.length;++He)(de[ke]==0||He<=Te||we.findIndex(gt=>gt==Le[He])!==-1)&&(We[He]=-1/0,qe[He]=-1/0);const st=(0,l.softmax)(We).map((He,gt)=>[He,gt]),Ze=(0,l.softmax)(qe).map((He,gt)=>[He,gt]);st[0][0]=0,Ze[0][0]=0;const ze=(0,i.product)(st,Ze).filter(He=>He[0][1]<=He[1][1]).map(He=>[He[0][1],He[1][1],He[0][0]*He[1][0]]).sort((He,gt)=>gt[2]-He[2]);for(let He=0;He<Math.min(ze.length,_e);++He){const[gt,dt,kt]=ze[He],ht=Le.slice(gt,dt+1),yr=this.tokenizer.decode(ht,{skip_special_tokens:!0});ce.push({answer:yr,score:kt})}}return _e===1?ce[0]:ce}}class C extends b{constructor(X){super(X)}async _call(X,{top_k:z=5}={}){const _e=this.tokenizer(X,{padding:!0,truncation:!0}),{logits:Ee}=await this.model(_e),Me=[],Ce=_e.input_ids.tolist();for(let ye=0;ye<Ce.length;++ye){const de=Ce[ye],we=de.findIndex(We=>We==this.tokenizer.mask_token_id);if(we===-1)throw Error(`Mask token (${this.tokenizer.mask_token}) not found in text.`);const ce=Ee[ye][we],ke=await(0,p.topk)(new p.Tensor("float32",(0,l.softmax)(ce.data),ce.dims),z),Le=ke[0].tolist(),Te=ke[1].tolist();Me.push(Te.map((We,qe)=>{const st=de.slice();return st[we]=We,{score:Le[qe],token:Number(We),token_str:this.tokenizer.decode([We]),sequence:this.tokenizer.decode(st,{skip_special_tokens:!0})}}))}return Array.isArray(X)?Me:Me[0]}}class x extends b{constructor(z){super(z);re(this,"_key","generated_text")}async _call(z,_e={}){Array.isArray(z)||(z=[z]),this.model.config.prefix&&(z=z.map(we=>this.model.config.prefix+we));const Ee=this.model.config.task_specific_params;Ee&&Ee[this.task]&&Ee[this.task].prefix&&(z=z.map(we=>Ee[this.task].prefix+we));const Me=this.tokenizer,Ce={padding:!0,truncation:!0};let ye;this instanceof T&&"_build_translation_inputs"in Me?ye=Me._build_translation_inputs(z,Ce,_e):ye=Me(z,Ce);const de=await this.model.generate({...ye,..._e});return Me.batch_decode(de,{skip_special_tokens:!0}).map(we=>({[this._key]:we}))}}class M extends x{constructor(z){super(z);re(this,"_key","summary_text")}}class T extends x{constructor(z){super(z);re(this,"_key","translation_text")}}function v($e){return Array.isArray($e)&&$e.every(X=>"role"in X&&"content"in X)}class P extends b{constructor(X){super(X)}async _call(X,z={}){let _e=!1,Ee=!1,Me=z.add_special_tokens??(this.tokenizer.add_bos_token||this.tokenizer.add_eos_token)??!1,Ce;if(typeof X=="string")Ce=X=[X];else if(Array.isArray(X)&&X.every(Te=>typeof Te=="string"))_e=!0,Ce=X;else{if(v(X))X=[X];else if(Array.isArray(X)&&X.every(v))_e=!0;else throw new Error("Input must be a string, an array of strings, a Chat, or an array of Chats");Ee=!0,Ce=X.map(Te=>this.tokenizer.apply_chat_template(Te,{tokenize:!1,add_generation_prompt:!0})),Me=!1}const ye=Ee?!1:z.return_full_text??!0;this.tokenizer.padding_side="left";const de=this.tokenizer(Ce,{add_special_tokens:Me,padding:!0,truncation:!0}),we=await this.model.generate({...de,...z}),ce=this.tokenizer.batch_decode(we,{skip_special_tokens:!0});let ke;!ye&&de.input_ids.dims.at(-1)>0&&(ke=this.tokenizer.batch_decode(de.input_ids,{skip_special_tokens:!0}).map(Te=>Te.length));const Le=Array.from({length:X.length},Te=>[]);for(let Te=0;Te<ce.length;++Te){const We=Math.floor(Te/we.dims[0]*X.length);ke&&(ce[Te]=ce[Te].slice(ke[We])),Le[We].push({generated_text:Ee?[...X[We],{role:"assistant",content:ce[Te]}]:ce[Te]})}return!_e&&Le.length===1?Le[0]:Le}}class F extends b{constructor(X){super(X),this.label2id=Object.fromEntries(Object.entries(this.model.config.label2id).map(([z,_e])=>[z.toLowerCase(),_e])),this.entailment_id=this.label2id.entailment,this.entailment_id===void 0&&(console.warn("Could not find 'entailment' in label2id mapping. Using 2 as entailment_id."),this.entailment_id=2),this.contradiction_id=this.label2id.contradiction??this.label2id.not_entailment,this.contradiction_id===void 0&&(console.warn("Could not find 'contradiction' in label2id mapping. Using 0 as contradiction_id."),this.contradiction_id=0)}async _call(X,z,{hypothesis_template:_e="This example is {}.",multi_label:Ee=!1}={}){const Me=Array.isArray(X);Me||(X=[X]),Array.isArray(z)||(z=[z]);const Ce=z.map(we=>_e.replace("{}",we)),ye=Ee||z.length===1,de=[];for(const we of X){const ce=[];for(const Te of Ce){const We=this.tokenizer(we,{text_pair:Te,padding:!0,truncation:!0}),qe=await this.model(We);ye?ce.push([qe.logits.data[this.contradiction_id],qe.logits.data[this.entailment_id]]):ce.push(qe.logits.data[this.entailment_id])}const Le=(ye?ce.map(Te=>(0,l.softmax)(Te)[1]):(0,l.softmax)(ce)).map((Te,We)=>[Te,We]).sort((Te,We)=>We[0]-Te[0]);de.push({sequence:we,labels:Le.map(Te=>z[Te[1]]),scores:Le.map(Te=>Te[0])})}return Me?de:de[0]}}class D extends b{constructor(X){super(X)}async _call(X,{pooling:z="none",normalize:_e=!1,quantize:Ee=!1,precision:Me="binary"}={}){const Ce=this.tokenizer(X,{padding:!0,truncation:!0}),ye=await this.model(Ce);let de=ye.last_hidden_state??ye.logits??ye.token_embeddings;switch(z){case"none":break;case"mean":de=(0,p.mean_pooling)(de,Ce.attention_mask);break;case"first_token":case"cls":de=de.slice(null,0);break;case"last_token":case"eos":de=de.slice(null,-1);break;default:throw Error(`Pooling method '${z}' not supported.`)}return _e&&(de=de.normalize(2,-1)),Ee&&(de=(0,p.quantize_embeddings)(de,Me)),de}}class K extends b{constructor(X){super(X)}async _call(X,{pool:z=null}={}){const _e=await u(X),{pixel_values:Ee}=await this.processor(_e),Me=await this.model({pixel_values:Ee});let Ce;if(z){if(!("pooler_output"in Me))throw Error("No pooled output was returned. Make sure the model has a 'pooler' layer when using the 'pool' option.");Ce=Me.pooler_output}else Ce=Me.last_hidden_state??Me.logits??Me.image_embeds;return Ce}}class U extends b{constructor(X){super(X)}async _call(X,{top_k:z=5}={}){const _e=this.processor.feature_extractor.config.sampling_rate,Ee=await _(X,_e),Me=this.model.config.id2label,Ce=[];for(const ye of Ee){const de=await this.processor(ye),ce=(await this.model(de)).logits[0],ke=await(0,p.topk)(new p.Tensor("float32",(0,l.softmax)(ce.data),ce.dims),z),Le=ke[0].tolist(),We=ke[1].tolist().map((qe,st)=>({label:Me?Me[qe]:`LABEL_${qe}`,score:Le[st]}));Ce.push(We)}return Array.isArray(X)?Ce:Ce[0]}}class j extends b{constructor(X){super(X)}async _call(X,z,{hypothesis_template:_e="This is a sound of {}."}={}){const Ee=!Array.isArray(X);Ee&&(X=[X]);const Me=z.map(ce=>_e.replace("{}",ce)),Ce=this.tokenizer(Me,{padding:!0,truncation:!0}),ye=this.processor.feature_extractor.config.sampling_rate,de=await _(X,ye),we=[];for(const ce of de){const ke=await this.processor(ce),Le=await this.model({...Ce,...ke}),Te=(0,l.softmax)(Le.logits_per_audio.data);we.push([...Te].map((We,qe)=>({score:We,label:z[qe]})))}return Ee?we[0]:we}}class ne extends b{constructor(X){super(X)}async _call(X,z={}){switch(this.model.config.model_type){case"whisper":case"lite-whisper":return this._call_whisper(X,z);case"wav2vec2":case"wav2vec2-bert":case"unispeech":case"unispeech-sat":case"hubert":case"parakeet_ctc":return this._call_wav2vec2(X,z);case"moonshine":return this._call_moonshine(X,z);default:throw new Error(`AutomaticSpeechRecognitionPipeline does not support model type '${this.model.config.model_type}'.`)}}async _call_wav2vec2(X,z){z.language&&console.warn('`language` parameter is not yet supported for `wav2vec2` models, defaulting to "English".'),z.task&&console.warn('`task` parameter is not yet supported for `wav2vec2` models, defaulting to "transcribe".');const _e=!Array.isArray(X);_e&&(X=[X]);const Ee=this.processor.feature_extractor.config.sampling_rate,Me=await _(X,Ee),Ce=[];for(const ye of Me){const de=await this.processor(ye),ce=(await this.model(de)).logits[0],ke=[];for(const Te of ce)ke.push((0,l.max)(Te.data)[1]);const Le=this.tokenizer.decode(ke,{skip_special_tokens:!0}).trim();Ce.push({text:Le})}return _e?Ce[0]:Ce}async _call_whisper(X,z){const _e=z.return_timestamps??!1,Ee=z.chunk_length_s??0,Me=z.force_full_sequences??!1;let Ce=z.stride_length_s??null;const ye={...z};_e==="word"&&(ye.return_token_timestamps=!0,ye.return_timestamps=!1);const de=!Array.isArray(X);de&&(X=[X]);const we=this.processor.feature_extractor.config.chunk_length/this.model.config.max_source_positions,ce=this.processor.feature_extractor.config.hop_length,ke=this.processor.feature_extractor.config.sampling_rate,Le=await _(X,ke),Te=[];for(const We of Le){let qe=[];if(Ee>0){if(Ce===null)Ce=Ee/6;else if(Ee<=Ce)throw Error("`chunk_length_s` must be larger than `stride_length_s`.");const ze=ke*Ee,He=ke*Ce,gt=ze-2*He;let dt=0;for(;;){const kt=dt+ze,ht=We.subarray(dt,kt),yr=await this.processor(ht),$r=dt===0,Vr=kt>=We.length;if(qe.push({stride:[ht.length,$r?0:He,Vr?0:He],input_features:yr.input_features,is_last:Vr}),Vr)break;dt+=gt}}else qe=[{stride:[We.length,0,0],input_features:(await this.processor(We)).input_features,is_last:!0}];for(const ze of qe){ye.num_frames=Math.floor(ze.stride[0]/ce);const He=await this.model.generate({inputs:ze.input_features,...ye});_e==="word"?(ze.tokens=He.sequences.tolist()[0],ze.token_timestamps=He.token_timestamps.tolist()[0].map(gt=>(0,l.round)(gt,2))):ze.tokens=He[0].tolist(),ze.stride=ze.stride.map(gt=>gt/ke)}const[st,Ze]=this.tokenizer._decode_asr(qe,{time_precision:we,return_timestamps:_e,force_full_sequences:Me});Te.push({text:st,...Ze})}return de?Te[0]:Te}async _call_moonshine(X,z){const _e=!Array.isArray(X);_e&&(X=[X]);const Ee=this.processor.feature_extractor.config.sampling_rate,Me=await _(X,Ee),Ce=[];for(const ye of Me){const de=await this.processor(ye),we=Math.floor(ye.length/Ee)*6,ce=await this.model.generate({max_new_tokens:we,...z,...de}),ke=this.processor.batch_decode(ce,{skip_special_tokens:!0})[0];Ce.push({text:ke})}return _e?Ce[0]:Ce}}class q extends b{constructor(X){super(X)}async _call(X,z={}){const _e=Array.isArray(X),Ee=await u(X),{pixel_values:Me}=await this.processor(Ee),Ce=[];for(const ye of Me){ye.dims=[1,...ye.dims];const de=await this.model.generate({inputs:ye,...z}),we=this.tokenizer.batch_decode(de,{skip_special_tokens:!0}).map(ce=>({generated_text:ce.trim()}));Ce.push(we)}return _e?Ce:Ce[0]}}class te extends b{constructor(X){super(X)}async _call(X,{top_k:z=5}={}){const _e=await u(X),{pixel_values:Ee}=await this.processor(_e),Me=await this.model({pixel_values:Ee}),Ce=this.model.config.id2label,ye=[];for(const de of Me.logits){const we=await(0,p.topk)(new p.Tensor("float32",(0,l.softmax)(de.data),de.dims),z),ce=we[0].tolist(),Le=we[1].tolist().map((Te,We)=>({label:Ce?Ce[Te]:`LABEL_${Te}`,score:ce[We]}));ye.push(Le)}return Array.isArray(X)?ye:ye[0]}}class Z extends b{constructor(X){super(X),this.subtasks_mapping={panoptic:"post_process_panoptic_segmentation",instance:"post_process_instance_segmentation",semantic:"post_process_semantic_segmentation"}}async _call(X,{threshold:z=.5,mask_threshold:_e=.5,overlap_mask_area_threshold:Ee=.8,label_ids_to_fuse:Me=null,target_sizes:Ce=null,subtask:ye=null}={}){if(Array.isArray(X)&&X.length!==1)throw Error("Image segmentation pipeline currently only supports a batch size of 1.");const we=await u(X),ce=we.map(ze=>[ze.height,ze.width]),ke=await this.processor(we),{inputNames:Le,outputNames:Te}=this.model.sessions.model;if(!Le.includes("pixel_values")){if(Le.length!==1)throw Error(`Expected a single input name, but got ${Le.length} inputs: ${Le}.`);const ze=Le[0];if(ze in ke)throw Error(`Input name ${ze} already exists in the inputs.`);ke[ze]=ke.pixel_values}const We=await this.model(ke);let qe=null;if(ye!==null)qe=this.subtasks_mapping[ye];else if(this.processor.image_processor){for(const[ze,He]of Object.entries(this.subtasks_mapping))if(He in this.processor.image_processor){qe=this.processor.image_processor[He].bind(this.processor.image_processor),ye=ze;break}}const st=this.model.config.id2label,Ze=[];if(ye)if(ye==="panoptic"||ye==="instance"){const ze=qe(We,z,_e,Ee,Me,Ce??ce)[0],He=ze.segmentation;for(const gt of ze.segments_info){const dt=new Uint8ClampedArray(He.data.length);for(let ht=0;ht<He.data.length;++ht)He.data[ht]===gt.id&&(dt[ht]=255);const kt=new d.RawImage(dt,He.dims[1],He.dims[0],1);Ze.push({score:gt.score,label:st[gt.label_id],mask:kt})}}else if(ye==="semantic"){const{segmentation:ze,labels:He}=qe(We,Ce??ce)[0];for(const gt of He){const dt=new Uint8ClampedArray(ze.data.length);for(let ht=0;ht<ze.data.length;++ht)ze.data[ht]===gt&&(dt[ht]=255);const kt=new d.RawImage(dt,ze.dims[1],ze.dims[0],1);Ze.push({score:null,label:st[gt],mask:kt})}}else throw Error(`Subtask ${ye} not supported.`);else{const He=We[Te[0]];for(let gt=0;gt<ce.length;++gt){const dt=ce[gt],kt=He[gt];kt.data.some(yr=>yr<-1e-5||yr>1+1e-5)&&kt.sigmoid_();const ht=await d.RawImage.fromTensor(kt.mul_(255).to("uint8")).resize(dt[1],dt[0]);Ze.push({label:null,score:null,mask:ht})}}return Ze}}class ae extends Z{constructor(X){super(X)}async _call(X,z={}){if(Array.isArray(X)&&X.length!==1)throw Error("Background removal pipeline currently only supports a batch size of 1.");const Ee=await u(X),Me=await super._call(X,z);return Ee.map((ye,de)=>{const we=ye.clone();return we.putAlpha(Me[de].mask),we})}}class he extends b{constructor(X){super(X)}async _call(X,z,{hypothesis_template:_e="This is a photo of {}"}={}){const Ee=Array.isArray(X),Me=await u(X),Ce=z.map(Le=>_e.replace("{}",Le)),ye=this.tokenizer(Ce,{padding:this.model.config.model_type==="siglip"?"max_length":!0,truncation:!0}),{pixel_values:de}=await this.processor(Me),we=await this.model({...ye,pixel_values:de}),ce=this.model.config.model_type==="siglip"?Le=>Le.sigmoid().data:Le=>(0,l.softmax)(Le.data),ke=[];for(const Le of we.logits_per_image){const We=[...ce(Le)].map((qe,st)=>({score:qe,label:z[st]}));We.sort((qe,st)=>st.score-qe.score),ke.push(We)}return Ee?ke:ke[0]}}class Q extends b{constructor(X){super(X)}async _call(X,{threshold:z=.9,percentage:_e=!1}={}){const Ee=Array.isArray(X);if(Ee&&X.length!==1)throw Error("Object detection pipeline currently only supports a batch size of 1.");const Me=await u(X),Ce=_e?null:Me.map(Te=>[Te.height,Te.width]),{pixel_values:ye,pixel_mask:de}=await this.processor(Me),we=await this.model({pixel_values:ye,pixel_mask:de}),ce=this.processor.image_processor.post_process_object_detection(we,z,Ce),ke=this.model.config.id2label,Le=ce.map(Te=>Te.boxes.map((We,qe)=>({score:Te.scores[qe],label:ke[Te.classes[qe]],box:f(We,!_e)})));return Ee?Le:Le[0]}}class B extends b{constructor(X){super(X)}async _call(X,z,{threshold:_e=.1,top_k:Ee=null,percentage:Me=!1}={}){const Ce=Array.isArray(X),ye=await u(X),de=this.tokenizer(z,{padding:!0,truncation:!0}),we=await this.processor(ye),ce=[];for(let ke=0;ke<ye.length;++ke){const Le=ye[ke],Te=Me?null:[[Le.height,Le.width]],We=we.pixel_values[ke].unsqueeze_(0),qe=await this.model({...de,pixel_values:We});let st;if("post_process_grounded_object_detection"in this.processor){const Ze=this.processor.post_process_grounded_object_detection(qe,de.input_ids,{box_threshold:_e,text_threshold:_e,target_sizes:Te})[0];st=Ze.boxes.map((ze,He)=>({score:Ze.scores[He],label:Ze.labels[He],box:f(ze,!Me)}))}else{const Ze=this.processor.image_processor.post_process_object_detection(qe,_e,Te,!0)[0];st=Ze.boxes.map((ze,He)=>({score:Ze.scores[He],label:z[Ze.classes[He]],box:f(ze,!Me)}))}st.sort((Ze,ze)=>ze.score-Ze.score),Ee!==null&&(st=st.slice(0,Ee)),ce.push(st)}return Ce?ce:ce[0]}}class O extends b{constructor(X){super(X)}async _call(X,z,_e={}){const Ee=(await u(X))[0],{pixel_values:Me}=await this.processor(Ee),Ce=`<s_docvqa><s_question>${z}</s_question><s_answer>`,ye=this.tokenizer(Ce,{add_special_tokens:!1,padding:!0,truncation:!0}).input_ids,de=await this.model.generate({inputs:Me,max_length:this.model.config.decoder.max_position_embeddings,decoder_input_ids:ye,..._e}),ce=this.tokenizer.batch_decode(de)[0].match(/<s_answer>(.*?)<\/s_answer>/);let ke=null;return ce&&ce.length>=2&&(ke=ce[1].trim()),[{answer:ke}]}}class W extends b{constructor(z){super(z);re(this,"DEFAULT_VOCODER_ID","Xenova/speecht5_hifigan");this.vocoder=z.vocoder??null}async _call(z,{speaker_embeddings:_e=null}={}){return this.processor?this._call_text_to_spectrogram(z,{speaker_embeddings:_e}):this._call_text_to_waveform(z)}async _call_text_to_waveform(z){const _e=this.tokenizer(z,{padding:!0,truncation:!0}),{waveform:Ee}=await this.model(_e),Me=this.model.config.sampling_rate;return new c.RawAudio(Ee.data,Me)}async _call_text_to_spectrogram(z,{speaker_embeddings:_e}){if(this.vocoder||(console.log("No vocoder specified, using default HifiGan vocoder."),this.vocoder=await o.AutoModel.from_pretrained(this.DEFAULT_VOCODER_ID,{dtype:"fp32"})),(typeof _e=="string"||_e instanceof URL)&&(_e=new Float32Array(await(await fetch(_e)).arrayBuffer())),_e instanceof Float32Array)_e=new p.Tensor("float32",_e,[1,_e.length]);else if(!(_e instanceof p.Tensor))throw new Error("Speaker embeddings must be a `Tensor`, `Float32Array`, `string`, or `URL`.");const{input_ids:Ee}=this.tokenizer(z,{padding:!0,truncation:!0}),{waveform:Me}=await this.model.generate_speech(Ee,_e,{vocoder:this.vocoder}),Ce=this.processor.feature_extractor.config.sampling_rate;return new c.RawAudio(Me.data,Ce)}}class N extends b{constructor(X){super(X)}async _call(X){const z=await u(X),_e=await this.processor(z),Ee=await this.model(_e),Me=[];for(const Ce of Ee.reconstruction){const ye=Ce.squeeze().clamp_(0,1).mul_(255).round_().to("uint8");Me.push(d.RawImage.fromTensor(ye))}return Me.length>1?Me:Me[0]}}class J extends b{constructor(X){super(X)}async _call(X){const z=await u(X),_e=await this.processor(z),{predicted_depth:Ee}=await this.model(_e),Me=[];for(let Ce=0;Ce<z.length;++Ce){const ye=Ee[Ce],[de,we]=ye.dims.slice(-2),[ce,ke]=z[Ce].size,Le=(await(0,p.interpolate_4d)(ye.view(1,1,de,we),{size:[ke,ce],mode:"bilinear"})).view(ke,ce),Te=Le.min().item(),We=Le.max().item(),qe=Le.sub(Te).div_(We-Te).mul_(255).to("uint8").unsqueeze(0),st=d.RawImage.fromTensor(qe);Me.push({predicted_depth:Le,depth:st})}return Me.length>1?Me:Me[0]}}const ie=Object.freeze({"text-classification":{tokenizer:s.AutoTokenizer,pipeline:A,model:o.AutoModelForSequenceClassification,default:{model:"Xenova/distilbert-base-uncased-finetuned-sst-2-english"},type:"text"},"token-classification":{tokenizer:s.AutoTokenizer,pipeline:g,model:o.AutoModelForTokenClassification,default:{model:"Xenova/bert-base-multilingual-cased-ner-hrl"},type:"text"},"question-answering":{tokenizer:s.AutoTokenizer,pipeline:y,model:o.AutoModelForQuestionAnswering,default:{model:"Xenova/distilbert-base-cased-distilled-squad"},type:"text"},"fill-mask":{tokenizer:s.AutoTokenizer,pipeline:C,model:o.AutoModelForMaskedLM,default:{model:"Xenova/bert-base-uncased"},type:"text"},summarization:{tokenizer:s.AutoTokenizer,pipeline:M,model:o.AutoModelForSeq2SeqLM,default:{model:"Xenova/distilbart-cnn-6-6"},type:"text"},translation:{tokenizer:s.AutoTokenizer,pipeline:T,model:o.AutoModelForSeq2SeqLM,default:{model:"Xenova/t5-small"},type:"text"},"text2text-generation":{tokenizer:s.AutoTokenizer,pipeline:x,model:o.AutoModelForSeq2SeqLM,default:{model:"Xenova/flan-t5-small"},type:"text"},"text-generation":{tokenizer:s.AutoTokenizer,pipeline:P,model:o.AutoModelForCausalLM,default:{model:"Xenova/gpt2"},type:"text"},"zero-shot-classification":{tokenizer:s.AutoTokenizer,pipeline:F,model:o.AutoModelForSequenceClassification,default:{model:"Xenova/distilbert-base-uncased-mnli"},type:"text"},"audio-classification":{pipeline:U,model:o.AutoModelForAudioClassification,processor:n.AutoProcessor,default:{model:"Xenova/wav2vec2-base-superb-ks"},type:"audio"},"zero-shot-audio-classification":{tokenizer:s.AutoTokenizer,pipeline:j,model:o.AutoModel,processor:n.AutoProcessor,default:{model:"Xenova/clap-htsat-unfused"},type:"multimodal"},"automatic-speech-recognition":{tokenizer:s.AutoTokenizer,pipeline:ne,model:[o.AutoModelForSpeechSeq2Seq,o.AutoModelForCTC],processor:n.AutoProcessor,default:{model:"Xenova/whisper-tiny.en"},type:"multimodal"},"text-to-audio":{tokenizer:s.AutoTokenizer,pipeline:W,model:[o.AutoModelForTextToWaveform,o.AutoModelForTextToSpectrogram],processor:[n.AutoProcessor,null],default:{model:"Xenova/speecht5_tts"},type:"text"},"image-to-text":{tokenizer:s.AutoTokenizer,pipeline:q,model:o.AutoModelForVision2Seq,processor:n.AutoProcessor,default:{model:"Xenova/vit-gpt2-image-captioning"},type:"multimodal"},"image-classification":{pipeline:te,model:o.AutoModelForImageClassification,processor:n.AutoProcessor,default:{model:"Xenova/vit-base-patch16-224"},type:"multimodal"},"image-segmentation":{pipeline:Z,model:[o.AutoModelForImageSegmentation,o.AutoModelForSemanticSegmentation,o.AutoModelForUniversalSegmentation],processor:n.AutoProcessor,default:{model:"Xenova/detr-resnet-50-panoptic"},type:"multimodal"},"background-removal":{pipeline:ae,model:[o.AutoModelForImageSegmentation,o.AutoModelForSemanticSegmentation,o.AutoModelForUniversalSegmentation],processor:n.AutoProcessor,default:{model:"Xenova/modnet"},type:"image"},"zero-shot-image-classification":{tokenizer:s.AutoTokenizer,pipeline:he,model:o.AutoModel,processor:n.AutoProcessor,default:{model:"Xenova/clip-vit-base-patch32"},type:"multimodal"},"object-detection":{pipeline:Q,model:o.AutoModelForObjectDetection,processor:n.AutoProcessor,default:{model:"Xenova/detr-resnet-50"},type:"multimodal"},"zero-shot-object-detection":{tokenizer:s.AutoTokenizer,pipeline:B,model:o.AutoModelForZeroShotObjectDetection,processor:n.AutoProcessor,default:{model:"Xenova/owlvit-base-patch32"},type:"multimodal"},"document-question-answering":{tokenizer:s.AutoTokenizer,pipeline:O,model:o.AutoModelForDocumentQuestionAnswering,processor:n.AutoProcessor,default:{model:"Xenova/donut-base-finetuned-docvqa"},type:"multimodal"},"image-to-image":{pipeline:N,model:o.AutoModelForImageToImage,processor:n.AutoProcessor,default:{model:"Xenova/swin2SR-classical-sr-x2-64"},type:"image"},"depth-estimation":{pipeline:J,model:o.AutoModelForDepthEstimation,processor:n.AutoProcessor,default:{model:"Xenova/dpt-large"},type:"image"},"feature-extraction":{tokenizer:s.AutoTokenizer,pipeline:D,model:o.AutoModel,default:{model:"Xenova/all-MiniLM-L6-v2"},type:"text"},"image-feature-extraction":{processor:n.AutoProcessor,pipeline:K,model:[o.AutoModelForImageFeatureExtraction,o.AutoModel],default:{model:"Xenova/vit-base-patch16-224-in21k"},type:"image"}}),me=Object.freeze({"sentiment-analysis":"text-classification",ner:"token-classification",asr:"automatic-speech-recognition","text-to-speech":"text-to-audio",embeddings:"feature-extraction"});async function Ae($e,X=null,{progress_callback:z=null,config:_e=null,cache_dir:Ee=null,local_files_only:Me=!1,revision:Ce="main",device:ye=null,dtype:de=null,subfolder:we="onnx",use_external_data_format:ce=null,model_file_name:ke=null,session_options:Le={}}={}){$e=me[$e]??$e;const Te=ie[$e.split("_",1)[0]];if(!Te)throw Error(`Unsupported pipeline: ${$e}. Must be one of [${Object.keys(ie)}]`);X||(X=Te.default.model,console.log(`No model specified. Using default model: "${X}".`));const We={progress_callback:z,config:_e,cache_dir:Ee,local_files_only:Me,revision:Ce,device:ye,dtype:de,subfolder:we,use_external_data_format:ce,model_file_name:ke,session_options:Le},qe=new Map([["tokenizer",Te.tokenizer],["model",Te.model],["processor",Te.processor]]),st=await Ve(qe,X,We);st.task=$e,(0,i.dispatchCallback)(z,{status:"ready",task:$e,model:X});const Ze=Te.pipeline;return new Ze(st)}async function Ve($e,X,z){const _e=Object.create(null),Ee=[];for(const[Me,Ce]of $e.entries()){if(!Ce)continue;let ye;Array.isArray(Ce)?ye=new Promise(async(de,we)=>{var ke,Le;let ce;for(const Te of Ce){if(Te===null){de(null);return}try{de(await Te.from_pretrained(X,z));return}catch(We){if((ke=We.message)!=null&&ke.includes("Unsupported model type"))ce=We;else if((Le=We.message)!=null&&Le.includes("Could not locate file"))ce=We;else{we(We);return}}}we(ce)}):ye=Ce.from_pretrained(X,z),_e[Me]=ye,Ee.push(ye)}await Promise.all(Ee);for(const[Me,Ce]of Object.entries(_e))_e[Me]=await Ce;return _e}},"./src/tokenizers.js":(e,r,t)=>{t.r(r),t.d(r,{AlbertTokenizer:()=>Ds,AutoTokenizer:()=>Cn,BartTokenizer:()=>tt,BertTokenizer:()=>Os,BlenderbotSmallTokenizer:()=>Qt,BlenderbotTokenizer:()=>rt,BloomTokenizer:()=>qr,CLIPTokenizer:()=>Gs,CamembertTokenizer:()=>De,CodeGenTokenizer:()=>fs,CodeLlamaTokenizer:()=>cr,CohereTokenizer:()=>Tn,ConvBertTokenizer:()=>Y,DebertaTokenizer:()=>$,DebertaV2Tokenizer:()=>ee,DistilBertTokenizer:()=>xe,ElectraTokenizer:()=>wt,Ernie4_5_Tokenizer:()=>Pn,EsmTokenizer:()=>zs,FalconTokenizer:()=>Ir,GPT2Tokenizer:()=>xt,GPTNeoXTokenizer:()=>Ls,GemmaTokenizer:()=>ms,Grok1Tokenizer:()=>Yr,HerbertTokenizer:()=>V,LlamaTokenizer:()=>kr,M2M100Tokenizer:()=>mr,MBart50Tokenizer:()=>qt,MBartTokenizer:()=>It,MPNetTokenizer:()=>hs,MarianTokenizer:()=>Ne,MgpstrTokenizer:()=>En,MobileBertTokenizer:()=>St,NllbTokenizer:()=>Cs,NougatTokenizer:()=>Ss,PreTrainedTokenizer:()=>ft,Qwen2Tokenizer:()=>vr,RoFormerTokenizer:()=>oe,RobertaTokenizer:()=>Wr,SiglipTokenizer:()=>Gr,SpeechT5Tokenizer:()=>Hs,SqueezeBertTokenizer:()=>Kt,T5Tokenizer:()=>pt,TokenizerModel:()=>K,VitsTokenizer:()=>ss,Wav2Vec2CTCTokenizer:()=>je,WhisperTokenizer:()=>ar,XLMRobertaTokenizer:()=>ps,XLMTokenizer:()=>nt,is_chinese_char:()=>C});var s=t("./src/utils/generic.js"),o=t("./src/utils/core.js"),n=t("./src/utils/hub.js"),a=t("./src/utils/maths.js"),i=t("./src/utils/tensor.js"),l=t("./src/utils/data-structures.js"),c=t("./node_modules/@huggingface/jinja/dist/index.js"),p=t("./src/models/whisper/common_whisper.js");async function d(ge,k){const G=await Promise.all([(0,n.getModelJSON)(ge,"tokenizer.json",!0,k),(0,n.getModelJSON)(ge,"tokenizer_config.json",!0,k)]);return k.legacy!==null&&(G[1].legacy=k.legacy),G}function u(ge,k){const G=[];let se=0;for(const ue of ge.matchAll(k)){const fe=ue[0];se<ue.index&&G.push(ge.slice(se,ue.index)),fe.length>0&&G.push(fe),se=ue.index+fe.length}return se<ge.length&&G.push(ge.slice(se)),G}function _(ge,k=!0){if(ge.Regex!==void 0){let G=ge.Regex.replace(/\\([#&~])/g,"$1");for(const[se,ue]of F)G=G.replaceAll(se,ue);return new RegExp(G,"gu")}else if(ge.String!==void 0){const G=(0,o.escapeRegExp)(ge.String);return new RegExp(k?G:`(${G})`,"gu")}else return console.warn("Unknown pattern type:",ge),null}function f(ge){return new Map(Object.entries(ge))}function b(ge){const k=ge.dims;switch(k.length){case 1:return ge.tolist();case 2:if(k[0]!==1)throw new Error("Unable to decode tensor with `batch size !== 1`. Use `tokenizer.batch_decode(...)` for batched inputs.");return ge.tolist()[0];default:throw new Error(`Expected tensor to have 1-2 dimensions, got ${k.length}.`)}}function A(ge){return ge.replace(/ \./g,".").replace(/ \?/g,"?").replace(/ \!/g,"!").replace(/ ,/g,",").replace(/ \' /g,"'").replace(/ n\'t/g,"n't").replace(/ \'m/g,"'m").replace(/ \'s/g,"'s").replace(/ \'ve/g,"'ve").replace(/ \'re/g,"'re")}function g(ge){return ge.replace(new RegExp("\\p{M}","gu"),"")}function y(ge){return g(ge.toLowerCase())}function C(ge){return ge>=19968&&ge<=40959||ge>=13312&&ge<=19903||ge>=131072&&ge<=173791||ge>=173824&&ge<=177983||ge>=177984&&ge<=178207||ge>=178208&&ge<=183983||ge>=63744&&ge<=64255||ge>=194560&&ge<=195103}function x(ge,k,G){const se=[];let ue=0;for(;ue<ge.length;){if(se.push(ge[ue]),(k.get(ge[ue])??G)!==G){++ue;continue}for(;++ue<ge.length&&(k.get(ge[ue])??G)===G;)k.get(se.at(-1))!==G&&(se[se.length-1]+=ge[ue])}return se}function M(ge){return ge.match(/\S+/g)||[]}const T="\\p{P}\\u0021-\\u002F\\u003A-\\u0040\\u005B-\\u0060\\u007B-\\u007E",v=new RegExp(`^[${T}]+$`,"gu"),P=".,!?…。，、।۔،",F=new Map([["(?i:'s|'t|'re|'ve|'m|'ll|'d)","(?:'([sS]|[tT]|[rR][eE]|[vV][eE]|[mM]|[lL][lL]|[dD]))"],["(?i:[sdmt]|ll|ve|re)","(?:[sS]|[dD]|[mM]|[tT]|[lL][lL]|[vV][eE]|[rR][eE])"],["[^\\r\\n\\p{L}\\p{N}]?+","[^\\r\\n\\p{L}\\p{N}]?"],["[^\\s\\p{L}\\p{N}]++","[^\\s\\p{L}\\p{N}]+"],[` ?[^(\\s|[${P}])]+`,` ?[^\\s${P}]+`]]);class D{constructor(k){this.content=k.content,this.id=k.id,this.single_word=k.single_word??!1,this.lstrip=k.lstrip??!1,this.rstrip=k.rstrip??!1,this.special=k.special??!1,this.normalized=k.normalized??null}}class K extends s.Callable{constructor(k){super(),this.config=k,this.vocab=[],this.tokens_to_ids=new Map,this.unk_token_id=void 0,this.unk_token=void 0,this.end_of_word_suffix=void 0,this.fuse_unk=this.config.fuse_unk??!1}static fromConfig(k,...G){switch(k.type){case"WordPiece":return new U(k);case"Unigram":return new j(k,...G);case"BPE":return new te(k);default:if(k.vocab)return Array.isArray(k.vocab)?new j(k,...G):Object.hasOwn(k,"continuing_subword_prefix")&&Object.hasOwn(k,"unk_token")?Object.hasOwn(k,"merges")?new te(k):new U(k):new Z(k,...G);throw new Error(`Unknown TokenizerModel type: ${k.type}`)}}_call(k){return k=this.encode(k),this.fuse_unk&&(k=x(k,this.tokens_to_ids,this.unk_token_id)),k}encode(k){throw Error("encode should be implemented in subclass.")}convert_tokens_to_ids(k){return k.map(G=>this.tokens_to_ids.get(G)??this.unk_token_id)}convert_ids_to_tokens(k){return k.map(G=>this.vocab[G]??this.unk_token)}}class U extends K{constructor(k){super(k),this.tokens_to_ids=f(k.vocab),this.unk_token_id=this.tokens_to_ids.get(k.unk_token),this.unk_token=k.unk_token,this.max_input_chars_per_word=k.max_input_chars_per_word??100,this.vocab=new Array(this.tokens_to_ids.size);for(const[G,se]of this.tokens_to_ids)this.vocab[se]=G}encode(k){const G=[];for(const se of k){const ue=[...se];if(ue.length>this.max_input_chars_per_word){G.push(this.unk_token);continue}let fe=!1,Se=0;const Ge=[];for(;Se<ue.length;){let Xe=ue.length,Qe=null;for(;Se<Xe;){let et=ue.slice(Se,Xe).join("");if(Se>0&&(et=this.config.continuing_subword_prefix+et),this.tokens_to_ids.has(et)){Qe=et;break}--Xe}if(Qe===null){fe=!0;break}Ge.push(Qe),Se=Xe}fe?G.push(this.unk_token):G.push(...Ge)}return G}}class j extends K{constructor(k,G){super(k);const se=k.vocab.length;this.vocab=new Array(se),this.scores=new Array(se);for(let ue=0;ue<se;++ue)[this.vocab[ue],this.scores[ue]]=k.vocab[ue];this.unk_token_id=k.unk_id,this.unk_token=this.vocab[k.unk_id],this.tokens_to_ids=new Map(this.vocab.map((ue,fe)=>[ue,fe])),this.bos_token=" ",this.bos_token_id=this.tokens_to_ids.get(this.bos_token),this.eos_token=G.eos_token,this.eos_token_id=this.tokens_to_ids.get(this.eos_token),this.unk_token=this.vocab[this.unk_token_id],this.minScore=(0,a.min)(this.scores)[0],this.unk_score=this.minScore-10,this.scores[this.unk_token_id]=this.unk_score,this.trie=new l.CharTrie,this.trie.extend(this.vocab),this.fuse_unk=!0}populateNodes(k){const G=k.chars,se=1;let ue=0;for(;ue<G.length;){let fe=!1;const Se=G.slice(ue).join(""),Ge=this.trie.commonPrefixSearch(Se);for(const Xe of Ge){const Qe=this.tokens_to_ids.get(Xe),et=this.scores[Qe],bt=(0,o.len)(Xe);k.insert(ue,bt,et,Qe),!fe&&bt===se&&(fe=!0)}fe||k.insert(ue,se,this.unk_score,this.unk_token_id),ue+=se}}tokenize(k){const G=new l.TokenLattice(k,this.bos_token_id,this.eos_token_id);return this.populateNodes(G),G.tokens()}encode(k){const G=[];for(const se of k){const ue=this.tokenize(se);G.push(...ue)}return G}}const ne=(()=>{const ge=[...Array.from({length:94},(ue,fe)=>fe+33),...Array.from({length:12},(ue,fe)=>fe+161),...Array.from({length:82},(ue,fe)=>fe+174)],k=ge.slice();let G=0;for(let ue=0;ue<256;++ue)ge.includes(ue)||(ge.push(ue),k.push(256+G),G+=1);const se=k.map(ue=>String.fromCharCode(ue));return Object.fromEntries(ge.map((ue,fe)=>[ue,se[fe]]))})(),q=(0,o.reverseDictionary)(ne);class te extends K{constructor(k){super(k),this.tokens_to_ids=f(k.vocab),this.unk_token_id=this.tokens_to_ids.get(k.unk_token),this.unk_token=k.unk_token,this.vocab=new Array(this.tokens_to_ids.size);for(const[se,ue]of this.tokens_to_ids)this.vocab[ue]=se;const G=Array.isArray(k.merges[0]);this.merges=G?k.merges:k.merges.map(se=>se.split(" ",2)),this.bpe_ranks=new Map(this.merges.map((se,ue)=>[JSON.stringify(se),ue])),this.end_of_word_suffix=k.end_of_word_suffix,this.continuing_subword_suffix=k.continuing_subword_suffix??null,this.byte_fallback=this.config.byte_fallback??!1,this.byte_fallback&&(this.text_encoder=new TextEncoder),this.ignore_merges=this.config.ignore_merges??!1,this.max_length_to_cache=256,this.cache_capacity=1e4,this.cache=new l.LRUCache(this.cache_capacity)}clear_cache(){this.cache.clear()}bpe(k){if(k.length===0)return[];const G=this.cache.get(k);if(G!==void 0)return G;const se=Array.from(k);this.end_of_word_suffix&&(se[se.length-1]+=this.end_of_word_suffix);let ue=[];if(se.length>1){const fe=new l.PriorityQueue((Xe,Qe)=>Xe.score<Qe.score);let Se={token:se[0],bias:0,prev:null,next:null},Ge=Se;for(let Xe=1;Xe<se.length;++Xe){const Qe={bias:Xe/se.length,token:se[Xe],prev:Ge,next:null};Ge.next=Qe,this._add_node(fe,Ge),Ge=Qe}for(;!fe.isEmpty();){const Xe=fe.pop();if(Xe.deleted||!Xe.next||Xe.next.deleted)continue;if(Xe.deleted=!0,Xe.next.deleted=!0,Xe.prev){const et={...Xe.prev};Xe.prev.deleted=!0,Xe.prev=et,et.prev?et.prev.next=et:Se=et}const Qe={token:Xe.token+Xe.next.token,bias:Xe.bias,prev:Xe.prev,next:Xe.next.next};Qe.prev?(Qe.prev.next=Qe,this._add_node(fe,Qe.prev)):Se=Qe,Qe.next&&(Qe.next.prev=Qe,this._add_node(fe,Qe))}for(let Xe=Se;Xe!==null;Xe=Xe.next)ue.push(Xe.token)}else ue=se;if(this.continuing_subword_suffix)for(let fe=0;fe<ue.length-1;++fe)ue[fe]+=this.continuing_subword_suffix;return k.length<this.max_length_to_cache&&this.cache.put(k,ue),ue}_add_node(k,G){const se=this.bpe_ranks.get(JSON.stringify([G.token,G.next.token]));se!==void 0&&(G.score=se+G.bias,k.push(G))}encode(k){const G=[];for(const se of k){if(this.ignore_merges&&this.tokens_to_ids.has(se)){G.push(se);continue}const ue=this.bpe(se);for(const fe of ue)if(this.tokens_to_ids.has(fe))G.push(fe);else if(this.byte_fallback){const Se=Array.from(this.text_encoder.encode(fe)).map(Ge=>`<0x${Ge.toString(16).toUpperCase().padStart(2,"0")}>`);Se.every(Ge=>this.tokens_to_ids.has(Ge))?G.push(...Se):G.push(this.unk_token)}else G.push(this.unk_token)}return G}}class Z extends K{constructor(k,G){super(k),this.tokens_to_ids=f(G.target_lang?k.vocab[G.target_lang]:k.vocab),this.bos_token=G.bos_token,this.bos_token_id=this.tokens_to_ids.get(this.bos_token),this.eos_token=G.eos_token,this.eos_token_id=this.tokens_to_ids.get(this.eos_token),this.pad_token=G.pad_token,this.pad_token_id=this.tokens_to_ids.get(this.pad_token),this.unk_token=G.unk_token,this.unk_token_id=this.tokens_to_ids.get(this.unk_token),this.vocab=new Array(this.tokens_to_ids.size);for(const[se,ue]of this.tokens_to_ids)this.vocab[ue]=se}encode(k){return k}}class ae extends s.Callable{constructor(k){super(),this.config=k}static fromConfig(k){if(k===null)return null;switch(k.type){case"BertNormalizer":return new $e(k);case"Precompiled":return new Vr(k);case"Sequence":return new Ve(k);case"Replace":return new he(k);case"NFC":return new B(k);case"NFD":return new O(k);case"NFKC":return new W(k);case"NFKD":return new N(k);case"Strip":return new J(k);case"StripAccents":return new ie(k);case"Lowercase":return new me(k);case"Prepend":return new Ae(k);default:throw new Error(`Unknown Normalizer type: ${k.type}`)}}normalize(k){throw Error("normalize should be implemented in subclass.")}_call(k){return this.normalize(k)}}class he extends ae{normalize(k){const G=_(this.config.pattern);return G===null?k:k.replaceAll(G,this.config.content)}}class Q extends ae{constructor(){super(...arguments);re(this,"form")}normalize(G){return G=G.normalize(this.form),G}}class B extends Q{constructor(){super(...arguments);re(this,"form","NFC")}}class O extends Q{constructor(){super(...arguments);re(this,"form","NFD")}}class W extends Q{constructor(){super(...arguments);re(this,"form","NFKC")}}class N extends Q{constructor(){super(...arguments);re(this,"form","NFKD")}}class J extends ae{normalize(k){return this.config.strip_left&&this.config.strip_right?k=k.trim():(this.config.strip_left&&(k=k.trimStart()),this.config.strip_right&&(k=k.trimEnd())),k}}class ie extends ae{normalize(k){return k=g(k),k}}class me extends ae{normalize(k){return k=k.toLowerCase(),k}}class Ae extends ae{normalize(k){return k=this.config.prepend+k,k}}class Ve extends ae{constructor(k){super(k),this.normalizers=k.normalizers.map(G=>ae.fromConfig(G))}normalize(k){return this.normalizers.reduce((G,se)=>se.normalize(G),k)}}class $e extends ae{_tokenize_chinese_chars(k){const G=[];for(let se=0;se<k.length;++se){const ue=k[se],fe=ue.charCodeAt(0);C(fe)?(G.push(" "),G.push(ue),G.push(" ")):G.push(ue)}return G.join("")}stripAccents(k){return k.normalize("NFD").replace(new RegExp("\\p{Mn}","gu"),"")}_is_control(k){switch(k){case"	":case`
`:case"\r":return!1;default:return new RegExp("^\\p{Cc}|\\p{Cf}|\\p{Co}|\\p{Cs}$","u").test(k)}}_clean_text(k){const G=[];for(const se of k){const ue=se.charCodeAt(0);ue===0||ue===65533||this._is_control(se)||(/^\s$/.test(se)?G.push(" "):G.push(se))}return G.join("")}normalize(k){return this.config.clean_text&&(k=this._clean_text(k)),this.config.handle_chinese_chars&&(k=this._tokenize_chinese_chars(k)),this.config.lowercase?(k=k.toLowerCase(),this.config.strip_accents!==!1&&(k=this.stripAccents(k))):this.config.strip_accents&&(k=this.stripAccents(k)),k}}class X extends s.Callable{static fromConfig(k){if(k===null)return null;switch(k.type){case"BertPreTokenizer":return new z(k);case"Sequence":return new Ur(k);case"Whitespace":return new sr(k);case"WhitespaceSplit":return new Ar(k);case"Metaspace":return new yr(k);case"ByteLevel":return new _e(k);case"Split":return new Ee(k);case"Punctuation":return new Me(k);case"Digits":return new Ce(k);case"Replace":return new rn(k);default:throw new Error(`Unknown PreTokenizer type: ${k.type}`)}}pre_tokenize_text(k,G){throw Error("pre_tokenize_text should be implemented in subclass.")}pre_tokenize(k,G){return(Array.isArray(k)?k.map(se=>this.pre_tokenize_text(se,G)):this.pre_tokenize_text(k,G)).flat()}_call(k,G){return this.pre_tokenize(k,G)}}class z extends X{constructor(k){super(),this.pattern=new RegExp(`[^\\s${T}]+|[${T}]`,"gu")}pre_tokenize_text(k,G){return k.trim().match(this.pattern)||[]}}class _e extends X{constructor(k){super(),this.config=k,this.add_prefix_space=this.config.add_prefix_space,this.trim_offsets=this.config.trim_offsets,this.use_regex=this.config.use_regex??!0,this.pattern=new RegExp("'s|'t|'re|'ve|'m|'ll|'d| ?\\p{L}+| ?\\p{N}+| ?[^\\s\\p{L}\\p{N}]+|\\s+(?!\\S)|\\s+","gu"),this.byte_encoder=ne,this.text_encoder=new TextEncoder}pre_tokenize_text(k,G){return this.add_prefix_space&&!k.startsWith(" ")&&(k=" "+k),(this.use_regex?k.match(this.pattern)||[]:[k]).map(ue=>Array.from(this.text_encoder.encode(ue),fe=>this.byte_encoder[fe]).join(""))}}class Ee extends X{constructor(k){super(),this.config=k,this.pattern=_(this.config.pattern,this.config.invert)}pre_tokenize_text(k,G){var se;return this.pattern===null?[]:this.config.invert?k.match(this.pattern)||[]:((se=this.config.behavior)==null?void 0:se.toLowerCase())==="removed"?k.split(this.pattern).filter(ue=>ue):u(k,this.pattern)}}class Me extends X{constructor(k){super(),this.config=k,this.pattern=new RegExp(`[^${T}]+|[${T}]+`,"gu")}pre_tokenize_text(k,G){return k.match(this.pattern)||[]}}class Ce extends X{constructor(k){super(),this.config=k;const G=`[^\\d]+|\\d${this.config.individual_digits?"":"+"}`;this.pattern=new RegExp(G,"gu")}pre_tokenize_text(k,G){return k.match(this.pattern)||[]}}class ye extends s.Callable{constructor(k){super(),this.config=k}static fromConfig(k){if(k===null)return null;switch(k.type){case"TemplateProcessing":return new ce(k);case"ByteLevel":return new ke(k);case"RobertaProcessing":return new we(k);case"BertProcessing":return new de(k);case"Sequence":return new Le(k);default:throw new Error(`Unknown PostProcessor type: ${k.type}`)}}post_process(k,...G){throw Error("post_process should be implemented in subclass.")}_call(k,...G){return this.post_process(k,...G)}}class de extends ye{constructor(k){super(k),this.cls=k.cls[0],this.sep=k.sep[0]}post_process(k,G=null,{add_special_tokens:se=!0}={}){se&&(k=(0,o.mergeArrays)([this.cls],k,[this.sep]));let ue=new Array(k.length).fill(0);if(G!==null){const fe=se&&this instanceof we?[this.sep]:[],Se=se?[this.sep]:[];k=(0,o.mergeArrays)(k,fe,G,Se),ue=(0,o.mergeArrays)(ue,new Array(G.length+fe.length+Se.length).fill(1))}return{tokens:k,token_type_ids:ue}}}class we extends de{}class ce extends ye{constructor(k){super(k),this.single=k.single,this.pair=k.pair}post_process(k,G=null,{add_special_tokens:se=!0}={}){const ue=G===null?this.single:this.pair;let fe=[],Se=[];for(const Ge of ue)"SpecialToken"in Ge?se&&(fe.push(Ge.SpecialToken.id),Se.push(Ge.SpecialToken.type_id)):"Sequence"in Ge&&(Ge.Sequence.id==="A"?(fe=(0,o.mergeArrays)(fe,k),Se=(0,o.mergeArrays)(Se,new Array(k.length).fill(Ge.Sequence.type_id))):Ge.Sequence.id==="B"&&(fe=(0,o.mergeArrays)(fe,G),Se=(0,o.mergeArrays)(Se,new Array(G.length).fill(Ge.Sequence.type_id))));return{tokens:fe,token_type_ids:Se}}}class ke extends ye{post_process(k,G=null){return G&&(k=(0,o.mergeArrays)(k,G)),{tokens:k}}}class Le extends ye{constructor(k){super(k),this.processors=k.processors.map(G=>ye.fromConfig(G))}post_process(k,G=null,se={}){let ue;for(const fe of this.processors)if(fe instanceof ke)k=fe.post_process(k).tokens,G&&(G=fe.post_process(G).tokens);else{const Se=fe.post_process(k,G,se);k=Se.tokens,ue=Se.token_type_ids}return{tokens:k,token_type_ids:ue}}}class Te extends s.Callable{constructor(k){super(),this.config=k,this.added_tokens=[],this.end_of_word_suffix=null,this.trim_offsets=k.trim_offsets}static fromConfig(k){if(k===null)return null;switch(k.type){case"WordPiece":return new ze(k);case"Metaspace":return new $r(k);case"ByteLevel":return new He(k);case"Replace":return new We(k);case"ByteFallback":return new qe(k);case"Fuse":return new st(k);case"Strip":return new Ze(k);case"Sequence":return new dt(k);case"CTC":return new gt(k);case"BPEDecoder":return new kt(k);default:throw new Error(`Unknown Decoder type: ${k.type}`)}}_call(k){return this.decode(k)}decode(k){return this.decode_chain(k).join("")}decode_chain(k){throw Error("`decode_chain` should be implemented in subclass.")}}class We extends Te{decode_chain(k){const G=_(this.config.pattern);return G===null?k:k.map(se=>se.replaceAll(G,this.config.content))}}class qe extends Te{constructor(k){super(k),this.text_decoder=new TextDecoder}decode_chain(k){const G=[];let se=[];for(const ue of k){let fe=null;if(ue.length===6&&ue.startsWith("<0x")&&ue.endsWith(">")){const Se=parseInt(ue.slice(3,5),16);isNaN(Se)||(fe=Se)}if(fe!==null)se.push(fe);else{if(se.length>0){const Se=this.text_decoder.decode(Uint8Array.from(se));G.push(Se),se=[]}G.push(ue)}}if(se.length>0){const ue=this.text_decoder.decode(Uint8Array.from(se));G.push(ue),se=[]}return G}}class st extends Te{decode_chain(k){return[k.join("")]}}class Ze extends Te{constructor(k){super(k),this.content=this.config.content,this.start=this.config.start,this.stop=this.config.stop}decode_chain(k){return k.map(G=>{let se=0;for(let fe=0;fe<this.start&&G[fe]===this.content;++fe){se=fe+1;continue}let ue=G.length;for(let fe=0;fe<this.stop;++fe){const Se=G.length-fe-1;if(G[Se]===this.content){ue=Se;continue}else break}return G.slice(se,ue)})}}class ze extends Te{constructor(k){super(k),this.cleanup=k.cleanup}decode_chain(k){return k.map((G,se)=>(se!==0&&(G.startsWith(this.config.prefix)?G=G.replace(this.config.prefix,""):G=" "+G),this.cleanup&&(G=A(G)),G))}}class He extends Te{constructor(k){super(k),this.byte_decoder=q,this.text_decoder=new TextDecoder("utf-8",{fatal:!1,ignoreBOM:!0}),this.end_of_word_suffix=null}convert_tokens_to_string(k){const G=k.join(""),se=new Uint8Array([...G].map(fe=>this.byte_decoder[fe]));return this.text_decoder.decode(se)}decode_chain(k){const G=[];let se=[];for(const ue of k)this.added_tokens.find(fe=>fe.content===ue)!==void 0?(se.length>0&&(G.push(this.convert_tokens_to_string(se)),se=[]),G.push(ue)):se.push(ue);return se.length>0&&G.push(this.convert_tokens_to_string(se)),G}}class gt extends Te{constructor(k){super(k),this.pad_token=this.config.pad_token,this.word_delimiter_token=this.config.word_delimiter_token,this.cleanup=this.config.cleanup}convert_tokens_to_string(k){if(k.length===0)return"";const G=[k[0]];for(let fe=1;fe<k.length;++fe)k[fe]!==G.at(-1)&&G.push(k[fe]);let ue=G.filter(fe=>fe!==this.pad_token).join("");return this.cleanup&&(ue=A(ue).replaceAll(this.word_delimiter_token," ").trim()),ue}decode_chain(k){return[this.convert_tokens_to_string(k)]}}class dt extends Te{constructor(k){super(k),this.decoders=k.decoders.map(G=>Te.fromConfig(G))}decode_chain(k){return this.decoders.reduce((G,se)=>se.decode_chain(G),k)}}class kt extends Te{constructor(k){super(k),this.suffix=this.config.suffix}decode_chain(k){return k.map((G,se)=>G.replaceAll(this.suffix,se===k.length-1?"":" "))}}class ht extends Te{decode_chain(k){let G="";for(let se=1;se<k.length;se+=2)G+=k[se];return[G]}}class yr extends X{constructor(k){super(),this.addPrefixSpace=k.add_prefix_space,this.replacement=k.replacement,this.strRep=k.str_rep||this.replacement,this.prepend_scheme=k.prepend_scheme??"always"}pre_tokenize_text(k,{section_index:G=void 0}={}){let se=k.replaceAll(" ",this.strRep);return this.addPrefixSpace&&!se.startsWith(this.replacement)&&(this.prepend_scheme==="always"||this.prepend_scheme==="first"&&G===0)&&(se=this.strRep+se),[se]}}class $r extends Te{constructor(k){super(k),this.addPrefixSpace=k.add_prefix_space,this.replacement=k.replacement}decode_chain(k){const G=[];for(let se=0;se<k.length;++se){let ue=k[se].replaceAll(this.replacement," ");this.addPrefixSpace&&se==0&&ue.startsWith(" ")&&(ue=ue.substring(1)),G.push(ue)}return G}}class Vr extends ae{constructor(k){super(k),this.charsmap=k.precompiled_charsmap}normalize(k){return k=k.replace(/[\u0001-\u0008\u000B\u000E-\u001F\u007F\u008F\u009F]/gm,""),k=k.replace(/[\u0009\u000A\u000C\u000D\u00A0\u1680\u2000-\u200F\u2028\u2029\u202F\u205F\u2581\u3000\uFEFF\uFFFD]/gm," "),k.includes("～")?k=k.split("～").map(se=>se.normalize("NFKC")).join("～"):k=k.normalize("NFKC"),k}}class Ur extends X{constructor(k){super(),this.tokenizers=k.pretokenizers.map(G=>X.fromConfig(G))}pre_tokenize_text(k,G){return this.tokenizers.reduce((se,ue)=>ue.pre_tokenize(se,G),[k])}}class sr extends X{constructor(k){super()}pre_tokenize_text(k,G){return k.match(/\w+|[^\w\s]+/g)||[]}}class Ar extends X{constructor(k){super()}pre_tokenize_text(k,G){return M(k)}}class rn extends X{constructor(k){super(),this.config=k,this.pattern=_(this.config.pattern),this.content=this.config.content}pre_tokenize_text(k,G){return this.pattern===null?[k]:[k.replaceAll(this.pattern,this.config.content)]}}const sn=["bos_token","eos_token","unk_token","sep_token","pad_token","cls_token","mask_token"];function nn(ge,k,G,se){for(const ue of Object.keys(ge)){const fe=k-ge[ue].length,Se=G(ue),Ge=new Array(fe).fill(Se);ge[ue]=se==="right"?(0,o.mergeArrays)(ge[ue],Ge):(0,o.mergeArrays)(Ge,ge[ue])}}function ds(ge,k){for(const G of Object.keys(ge))ge[G].length=k}class ft extends s.Callable{constructor(G,se){super();re(this,"return_token_type_ids",!1);re(this,"padding_side","right");this.config=se,this.normalizer=ae.fromConfig(G.normalizer),this.pre_tokenizer=X.fromConfig(G.pre_tokenizer),this.model=K.fromConfig(G.model,se),this.post_processor=ye.fromConfig(G.post_processor),this.decoder=Te.fromConfig(G.decoder),this.special_tokens=[],this.all_special_ids=[],this.added_tokens=[];for(const ue of G.added_tokens){const fe=new D(ue);this.added_tokens.push(fe),this.model.tokens_to_ids.set(fe.content,fe.id),this.model.vocab[fe.id]=fe.content,fe.special&&(this.special_tokens.push(fe.content),this.all_special_ids.push(fe.id))}if(this.additional_special_tokens=se.additional_special_tokens??[],this.special_tokens.push(...this.additional_special_tokens),this.special_tokens=[...new Set(this.special_tokens)],this.decoder&&(this.decoder.added_tokens=this.added_tokens,this.decoder.end_of_word_suffix=this.model.end_of_word_suffix),this.added_tokens_splitter=new l.DictionarySplitter(this.added_tokens.map(ue=>ue.content)),this.added_tokens_map=new Map(this.added_tokens.map(ue=>[ue.content,ue])),this.mask_token=this.getToken("mask_token"),this.mask_token_id=this.model.tokens_to_ids.get(this.mask_token),this.pad_token=this.getToken("pad_token","eos_token"),this.pad_token_id=this.model.tokens_to_ids.get(this.pad_token),this.sep_token=this.getToken("sep_token"),this.sep_token_id=this.model.tokens_to_ids.get(this.sep_token),this.unk_token=this.getToken("unk_token"),this.unk_token_id=this.model.tokens_to_ids.get(this.unk_token),this.bos_token=this.getToken("bos_token"),this.bos_token_id=this.model.tokens_to_ids.get(this.bos_token),this.eos_token=this.getToken("eos_token"),this.eos_token_id=this.model.tokens_to_ids.get(this.eos_token),this.model_max_length=se.model_max_length,this.remove_space=se.remove_space,this.clean_up_tokenization_spaces=se.clean_up_tokenization_spaces??!0,this.do_lowercase_and_remove_accent=se.do_lowercase_and_remove_accent??!1,se.padding_side&&(this.padding_side=se.padding_side),this.add_bos_token=se.add_bos_token,this.add_eos_token=se.add_eos_token,this.legacy=!1,this.chat_template=se.chat_template??null,Array.isArray(this.chat_template)){const ue=Object.create(null);for(const{name:fe,template:Se}of this.chat_template){if(typeof fe!="string"||typeof Se!="string")throw new Error('Chat template must be a list of objects with "name" and "template" properties');ue[fe]=Se}this.chat_template=ue}this._compiled_template_cache=new Map}getToken(...G){for(const se of G){const ue=this.config[se];if(ue)if(typeof ue=="object"){if(ue.__type==="AddedToken")return ue.content;throw Error(`Unknown token: ${ue}`)}else return ue}return null}static async from_pretrained(G,{progress_callback:se=null,config:ue=null,cache_dir:fe=null,local_files_only:Se=!1,revision:Ge="main",legacy:Xe=null}={}){const Qe=await d(G,{progress_callback:se,config:ue,cache_dir:fe,local_files_only:Se,revision:Ge,legacy:Xe});return new this(...Qe)}_call(G,{text_pair:se=null,add_special_tokens:ue=!0,padding:fe=!1,truncation:Se=null,max_length:Ge=null,return_tensor:Xe=!0,return_token_type_ids:Qe=null}={}){const et=Array.isArray(G);let bt;if(et){if(G.length===0)throw Error("text array must be non-empty");if(se!==null){if(Array.isArray(se)){if(G.length!==se.length)throw Error("text and text_pair must have the same length")}else throw Error("text_pair must also be an array");bt=G.map((Lt,Zt)=>this._encode_plus(Lt,{text_pair:se[Zt],add_special_tokens:ue,return_token_type_ids:Qe}))}else bt=G.map(Lt=>this._encode_plus(Lt,{add_special_tokens:ue,return_token_type_ids:Qe}))}else{if(G==null)throw Error("text may not be null or undefined");if(Array.isArray(se))throw Error("When specifying `text_pair`, since `text` is a string, `text_pair` must also be a string (i.e., not an array).");bt=[this._encode_plus(G,{text_pair:se,add_special_tokens:ue,return_token_type_ids:Qe})]}if(Ge===null?Ge=this.model_max_length:Se===null&&(fe===!0?(console.warn("`max_length` is ignored when `padding: true` and there is no truncation strategy. To pad to max length, use `padding: 'max_length'`."),Ge=this.model_max_length):fe===!1&&(console.warn("Truncation was not explicitly activated but `max_length` is provided a specific value, please use `truncation: true` to explicitly truncate examples to max length."),Se=!0)),fe===!0&&(Ge=Math.min((0,a.max)(bt.map(Lt=>Lt.input_ids.length))[0],Ge??1/0)),Ge=Math.min(Ge,this.model_max_length??1/0),fe||Se)for(let Lt=0;Lt<bt.length;++Lt)bt[Lt].input_ids.length!==Ge&&(bt[Lt].input_ids.length>Ge?Se&&ds(bt[Lt],Ge):fe&&nn(bt[Lt],Ge,Zt=>Zt==="input_ids"?this.pad_token_id:0,this.padding_side));const Rt={};if(Xe){if(!(fe&&Se)&&bt.some(Zt=>{var Gt;for(const _r of Object.keys(Zt))if(Zt[_r].length!==((Gt=bt[0][_r])==null?void 0:Gt.length))return!0;return!1}))throw Error("Unable to create tensor, you should probably activate truncation and/or padding with 'padding=true' and 'truncation=true' to have batched tensors with the same length.");const Lt=[bt.length,bt[0].input_ids.length];for(const Zt of Object.keys(bt[0]))Rt[Zt]=new i.Tensor("int64",BigInt64Array.from(bt.flatMap(Gt=>Gt[Zt]).map(BigInt)),Lt)}else{for(const Lt of Object.keys(bt[0]))Rt[Lt]=bt.map(Zt=>Zt[Lt]);if(!et)for(const Lt of Object.keys(Rt))Rt[Lt]=Rt[Lt][0]}return Rt}_encode_text(G){if(G===null)return null;const se=this.added_tokens_splitter.split(G);for(let fe=0;fe<se.length;++fe){const Se=this.added_tokens_map.get(se[fe]);Se&&(Se.lstrip&&fe>0&&(se[fe-1]=se[fe-1].trimEnd()),Se.rstrip&&fe<se.length-1&&(se[fe+1]=se[fe+1].trimStart()))}return se.flatMap((fe,Se)=>{if(fe.length===0)return[];if(this.added_tokens_map.has(fe))return[fe];if(this.remove_space===!0&&(fe=fe.trim().split(/\s+/).join(" ")),this.do_lowercase_and_remove_accent&&(fe=y(fe)),this.normalizer!==null&&(fe=this.normalizer(fe)),fe.length===0)return[];const Ge=this.pre_tokenizer!==null?this.pre_tokenizer(fe,{section_index:Se}):[fe];return this.model(Ge)})}_encode_plus(G,{text_pair:se=null,add_special_tokens:ue=!0,return_token_type_ids:fe=null}={}){const{tokens:Se,token_type_ids:Ge}=this._tokenize_helper(G,{pair:se,add_special_tokens:ue}),Xe=this.model.convert_tokens_to_ids(Se),Qe={input_ids:Xe,attention_mask:new Array(Xe.length).fill(1)};return(fe??this.return_token_type_ids)&&Ge&&(Qe.token_type_ids=Ge),Qe}_tokenize_helper(G,{pair:se=null,add_special_tokens:ue=!1}={}){const fe=this._encode_text(G),Se=this._encode_text(se);return this.post_processor?this.post_processor(fe,Se,{add_special_tokens:ue}):{tokens:(0,o.mergeArrays)(fe??[],Se??[])}}tokenize(G,{pair:se=null,add_special_tokens:ue=!1}={}){return this._tokenize_helper(G,{pair:se,add_special_tokens:ue}).tokens}encode(G,{text_pair:se=null,add_special_tokens:ue=!0,return_token_type_ids:fe=null}={}){return this._encode_plus(G,{text_pair:se,add_special_tokens:ue,return_token_type_ids:fe}).input_ids}batch_decode(G,se={}){return G instanceof i.Tensor&&(G=G.tolist()),G.map(ue=>this.decode(ue,se))}decode(G,se={}){if(G instanceof i.Tensor&&(G=b(G)),!Array.isArray(G)||G.length===0||!(0,o.isIntegralNumber)(G[0]))throw Error("token_ids must be a non-empty array of integers.");return this.decode_single(G,se)}decode_single(G,{skip_special_tokens:se=!1,clean_up_tokenization_spaces:ue=null}){let fe=this.model.convert_ids_to_tokens(G);se&&(fe=fe.filter(Ge=>!this.special_tokens.includes(Ge)));let Se=this.decoder?this.decoder(fe):fe.join(" ");return this.decoder&&this.decoder.end_of_word_suffix&&(Se=Se.replaceAll(this.decoder.end_of_word_suffix," "),se&&(Se=Se.trim())),(ue??this.clean_up_tokenization_spaces)&&(Se=A(Se)),Se}get_chat_template({chat_template:G=null,tools:se=null}={}){if(this.chat_template&&typeof this.chat_template=="object"){const ue=this.chat_template;if(G!==null&&Object.hasOwn(ue,G))G=ue[G];else if(G===null)if(se!==null&&"tool_use"in ue)G=ue.tool_use;else if("default"in ue)G=ue.default;else throw Error(`This model has multiple chat templates with no default specified! Please either pass a chat template or the name of the template you wish to use to the 'chat_template' argument. Available template names are ${Object.keys(ue).sort()}.`)}else if(G===null)if(this.chat_template)G=this.chat_template;else throw Error("Cannot use apply_chat_template() because tokenizer.chat_template is not set and no template argument was passed! For information about writing templates and setting the tokenizer.chat_template attribute, please see the documentation at https://huggingface.co/docs/transformers/main/en/chat_templating");return G}apply_chat_template(G,{tools:se=null,documents:ue=null,chat_template:fe=null,add_generation_prompt:Se=!1,tokenize:Ge=!0,padding:Xe=!1,truncation:Qe=!1,max_length:et=null,return_tensor:bt=!0,return_dict:Rt=!1,tokenizer_kwargs:Lt={},...Zt}={}){if(fe=this.get_chat_template({chat_template:fe,tools:se}),typeof fe!="string")throw Error(`chat_template must be a string, but got ${typeof fe}`);let Gt=this._compiled_template_cache.get(fe);Gt===void 0&&(Gt=new c.Template(fe),this._compiled_template_cache.set(fe,Gt));const _r=Object.create(null);for(const dr of sn){const wr=this.getToken(dr);wr&&(_r[dr]=wr)}const gr=Gt.render({messages:G,add_generation_prompt:Se,tools:se,documents:ue,..._r,...Zt});if(Ge){const dr=this._call(gr,{add_special_tokens:!1,padding:Xe,truncation:Qe,max_length:et,return_tensor:bt,...Lt});return Rt?dr:dr.input_ids}return gr}}class Os extends ft{constructor(){super(...arguments);re(this,"return_token_type_ids",!0)}}class Ds extends ft{constructor(){super(...arguments);re(this,"return_token_type_ids",!0)}}class St extends ft{constructor(){super(...arguments);re(this,"return_token_type_ids",!0)}}class Kt extends ft{constructor(){super(...arguments);re(this,"return_token_type_ids",!0)}}class $ extends ft{constructor(){super(...arguments);re(this,"return_token_type_ids",!0)}}class ee extends ft{constructor(){super(...arguments);re(this,"return_token_type_ids",!0)}}class V extends ft{constructor(){super(...arguments);re(this,"return_token_type_ids",!0)}}class Y extends ft{constructor(){super(...arguments);re(this,"return_token_type_ids",!0)}}class oe extends ft{constructor(){super(...arguments);re(this,"return_token_type_ids",!0)}}class xe extends ft{}class De extends ft{}class nt extends ft{constructor(G,se){super(G,se);re(this,"return_token_type_ids",!0);console.warn('WARNING: `XLMTokenizer` is not yet supported by Hugging Face\'s "fast" tokenizers library. Therefore, you may experience slightly inaccurate results.')}}class wt extends ft{constructor(){super(...arguments);re(this,"return_token_type_ids",!0)}}class pt extends ft{}class xt extends ft{}class tt extends ft{}class It extends ft{constructor(k,G){super(k,G),this.languageRegex=/^[a-z]{2}_[A-Z]{2}$/,this.language_codes=this.special_tokens.filter(se=>this.languageRegex.test(se)),this.lang_to_token=se=>se}_build_translation_inputs(k,G,se){return Rr(this,k,G,se)}}class qt extends It{}class Wr extends ft{}class qr extends ft{}const nr="▁";class kr extends ft{constructor(G,se){super(G,se);re(this,"padding_side","left");this.legacy=se.legacy??!0,this.legacy||(this.normalizer=null,this.pre_tokenizer=new yr({replacement:nr,add_prefix_space:!0,prepend_scheme:"first"}))}_encode_text(G){if(G===null)return null;if(this.legacy||G.length===0)return super._encode_text(G);let se=super._encode_text(nr+G.replaceAll(nr," "));return se.length>1&&se[0]===nr&&this.special_tokens.includes(se[1])&&(se=se.slice(1)),se}}class cr extends ft{}class ps extends ft{}class hs extends ft{}class Ir extends ft{}class Ls extends ft{}class zs extends ft{}class vr extends ft{}class ms extends ft{}class Yr extends ft{}function Rr(ge,k,G,se){if(!("language_codes"in ge)||!Array.isArray(ge.language_codes))throw new Error("Tokenizer must have `language_codes` attribute set and it should be an array of language ids.");if(!("languageRegex"in ge)||!(ge.languageRegex instanceof RegExp))throw new Error("Tokenizer must have `languageRegex` attribute set and it should be a regular expression.");if(!("lang_to_token"in ge)||typeof ge.lang_to_token!="function")throw new Error("Tokenizer must have `lang_to_token` attribute set and it should be a function.");const ue=se.src_lang,fe=se.tgt_lang;if(!ge.language_codes.includes(fe))throw new Error(`Target language code "${fe}" is not valid. Must be one of: {${ge.language_codes.join(", ")}}`);if(ue!==void 0){if(!ge.language_codes.includes(ue))throw new Error(`Source language code "${ue}" is not valid. Must be one of: {${ge.language_codes.join(", ")}}`);for(const Se of ge.post_processor.config.single)if("SpecialToken"in Se&&ge.languageRegex.test(Se.SpecialToken.id)){Se.SpecialToken.id=ge.lang_to_token(ue);break}}return se.forced_bos_token_id=ge.model.convert_tokens_to_ids([ge.lang_to_token(fe)])[0],ge._call(k,G)}class Cs extends ft{constructor(k,G){super(k,G),this.languageRegex=/^[a-z]{3}_[A-Z][a-z]{3}$/,this.language_codes=this.special_tokens.filter(se=>this.languageRegex.test(se)),this.lang_to_token=se=>se}_build_translation_inputs(k,G,se){return Rr(this,k,G,se)}}class mr extends ft{constructor(k,G){super(k,G),this.languageRegex=/^__[a-z]{2,3}__$/,this.language_codes=this.special_tokens.filter(se=>this.languageRegex.test(se)).map(se=>se.slice(2,-2)),this.lang_to_token=se=>`__${se}__`}_build_translation_inputs(k,G,se){return Rr(this,k,G,se)}}class ar extends ft{get timestamp_begin(){return this.model.convert_tokens_to_ids(["<|notimestamps|>"])[0]+1}_decode_asr(k,{return_timestamps:G=!1,return_language:se=!1,time_precision:ue=null,force_full_sequences:fe=!0}={}){if(ue===null)throw Error("Must specify time_precision");let Se=null;const Ge=G==="word";function Xe(){return{language:Se,timestamp:[null,null],text:""}}const Qe=[];let et=Xe(),bt=0;const Rt=this.timestamp_begin,Zt=Rt+1500;let Gt=[],_r=[],gr=!1,dr=null;const wr=new Set(this.all_special_ids);for(const Yt of k){const Mr=Yt.tokens,Fr=Ge?Yt.token_timestamps:null;let Zr=null,$s=Rt;if("stride"in Yt){const[ir,pr,fr]=Yt.stride;if(bt-=pr,dr=ir-fr,pr&&($s=pr/ue+Rt),fr)for(let er=Mr.length-1;er>=0;--er){const Qr=Number(Mr[er]);if(Qr>=Rt){if(Zr!==null&&(Qr-Rt)*ue<dr)break;Zr=Qr}}}let Or=[],_s=[];for(let ir=0;ir<Mr.length;++ir){const pr=Number(Mr[ir]);if(wr.has(pr)){const fr=this.decode([pr]),er=p.WHISPER_LANGUAGE_MAPPING.get(fr.slice(2,-2));if(er!==void 0){if(Se!==null&&er!==Se&&!G){Gt.push(Or);const Qr=this.findLongestCommonSequence(Gt)[0],Ks=this.decode(Qr);et.text=Ks,Qe.push(et),Gt=[],Or=[],et=Xe()}Se=et.language=er}}else if(pr>=Rt&&pr<=Zt){const fr=(pr-Rt)*ue+bt,er=(0,a.round)(fr,2);if(Zr!==null&&pr>=Zr)gr=!0;else if(gr||Gt.length>0&&pr<$s)gr=!1;else if(et.timestamp[0]===null)et.timestamp[0]=er;else if(er!==et.timestamp[0]){et.timestamp[1]=er,Gt.push(Or),Ge&&_r.push(_s);const[Qr,Ks]=this.findLongestCommonSequence(Gt,_r),Rs=this.decode(Qr);et.text=Rs,Ge&&(et.words=this.collateWordTimestamps(Qr,Ks,Se)),Qe.push(et),Gt=[],Or=[],_r=[],_s=[],et=Xe()}}else if(Or.push(pr),Ge){let fr=(0,a.round)(Fr[ir]+bt,2),er;if(ir+1<Fr.length){er=(0,a.round)(Fr[ir+1]+bt,2);const Qr=this.decode([pr]);v.test(Qr)&&(er=(0,a.round)(Math.min(fr+ue,er),2))}else er=null;_s.push([fr,er])}}if("stride"in Yt){const[ir,pr,fr]=Yt.stride;bt+=ir-fr}Or.length>0?(Gt.push(Or),Ge&&_r.push(_s)):Gt.every(ir=>ir.length===0)&&(et=Xe(),Gt=[],Or=[],_r=[],_s=[])}if(Gt.length>0){if(fe&&G)throw new Error("Whisper did not predict an ending timestamp, which can happen if audio is cut off in the middle of a word. Also make sure WhisperTimeStampLogitsProcessor was used during generation.");const[Yt,Mr]=this.findLongestCommonSequence(Gt,_r),Fr=this.decode(Yt);et.text=Fr,Ge&&(et.words=this.collateWordTimestamps(Yt,Mr,Se)),Qe.push(et)}let xr=Object.create(null);const ns=Qe.map(Yt=>Yt.text).join("");if(G||se){for(let Yt=0;Yt<Qe.length;++Yt){const Mr=Qe[Yt];G||delete Mr.timestamp,se||delete Mr.language}if(Ge){const Yt=[];for(const Mr of Qe)for(const Fr of Mr.words)Yt.push(Fr);xr={chunks:Yt}}else xr={chunks:Qe}}return[ns,xr]}findLongestCommonSequence(k,G=null){let se=k[0],ue=se.length,fe=[];const Se=Array.isArray(G)&&G.length>0;let Ge=Se?[]:null,Xe=Se?G[0]:null;for(let Qe=1;Qe<k.length;++Qe){const et=k[Qe];let bt=0,Rt=[ue,ue,0,0];const Lt=et.length;for(let xr=1;xr<ue+Lt;++xr){const ns=Math.max(0,ue-xr),Yt=Math.min(ue,ue+Lt-xr),Mr=se.slice(ns,Yt),Fr=Math.max(0,xr-ue),Zr=Math.min(Lt,xr),$s=et.slice(Fr,Zr);if(Mr.length!==$s.length)throw new Error("There is a bug within whisper `decode_asr` function, please report it. Dropping to prevent bad inference.");let Or;Se?Or=Mr.filter((pr,fr)=>pr===$s[fr]&&Xe[ns+fr]<=G[Qe][Fr+fr]).length:Or=Mr.filter((pr,fr)=>pr===$s[fr]).length;const _s=xr/1e4,ir=Or/xr+_s;Or>1&&ir>bt&&(bt=ir,Rt=[ns,Yt,Fr,Zr])}const[Zt,Gt,_r,gr]=Rt,dr=Math.floor((Gt+Zt)/2),wr=Math.floor((gr+_r)/2);fe.push(...se.slice(0,dr)),se=et.slice(wr),ue=se.length,Se&&(Ge.push(...Xe.slice(0,dr)),Xe=G[Qe].slice(wr))}return fe.push(...se),Se?(Ge.push(...Xe),[fe,Ge]):[fe,[]]}collateWordTimestamps(k,G,se){const[ue,fe,Se]=this.combineTokensIntoWords(k,se),Ge=[];for(let Xe=0;Xe<ue.length;++Xe){const Qe=Se[Xe];Ge.push({text:ue[Xe],timestamp:[G[Qe.at(0)][0],G[Qe.at(-1)][1]]})}return Ge}combineTokensIntoWords(k,G,se=`"'“¡¿([{-`,ue=`"'.。,，!！?？:：”)]}、`){G=G??"english";let fe,Se,Ge;return["chinese","japanese","thai","lao","myanmar"].includes(G)?[fe,Se,Ge]=this.splitTokensOnUnicode(k):[fe,Se,Ge]=this.splitTokensOnSpaces(k),this.mergePunctuations(fe,Se,Ge,se,ue)}decode(k,G){let se;return G!=null&&G.decode_with_timestamps?(k instanceof i.Tensor&&(k=b(k)),se=this.decodeWithTimestamps(k,G)):se=super.decode(k,G),se}decodeWithTimestamps(k,G){const se=(G==null?void 0:G.time_precision)??.02,ue=Array.from(this.all_special_ids).at(-1)+1;let fe=[[]];for(let Se of k)if(Se=Number(Se),Se>=ue){const Ge=((Se-ue)*se).toFixed(2);fe.push(`<|${Ge}|>`),fe.push([])}else fe[fe.length-1].push(Se);return fe=fe.map(Se=>typeof Se=="string"?Se:super.decode(Se,G)),fe.join("")}splitTokensOnUnicode(k){const G=this.decode(k,{decode_with_timestamps:!0}),se="�",ue=[],fe=[],Se=[];let Ge=[],Xe=[],Qe=0;for(let et=0;et<k.length;++et){const bt=k[et];Ge.push(bt),Xe.push(et);const Rt=this.decode(Ge,{decode_with_timestamps:!0});(!Rt.includes(se)||G[Qe+Rt.indexOf(se)]===se)&&(ue.push(Rt),fe.push(Ge),Se.push(Xe),Ge=[],Xe=[],Qe+=Rt.length)}return[ue,fe,Se]}splitTokensOnSpaces(k){const[G,se,ue]=this.splitTokensOnUnicode(k),fe=[],Se=[],Ge=[],Xe=new RegExp(`^[${T}]$`,"gu");for(let Qe=0;Qe<G.length;++Qe){const et=G[Qe],bt=se[Qe],Rt=ue[Qe],Lt=bt[0]>=this.model.tokens_to_ids.get("<|endoftext|>"),Zt=et.startsWith(" "),Gt=et.trim(),_r=Xe.test(Gt);if(Lt||Zt||_r||fe.length===0)fe.push(et),Se.push(bt),Ge.push(Rt);else{const gr=fe.length-1;fe[gr]+=et,Se[gr].push(...bt),Ge[gr].push(...Rt)}}return[fe,Se,Ge]}mergePunctuations(k,G,se,ue,fe){const Se=structuredClone(k),Ge=structuredClone(G),Xe=structuredClone(se);let Qe=Se.length-2,et=Se.length-1;for(;Qe>=0;)Se[Qe].startsWith(" ")&&ue.includes(Se[Qe].trim())?(Se[et]=Se[Qe]+Se[et],Ge[et]=(0,o.mergeArrays)(Ge[Qe],Ge[et]),Xe[et]=(0,o.mergeArrays)(Xe[Qe],Xe[et]),Se[Qe]="",Ge[Qe]=[],Xe[Qe]=[]):et=Qe,--Qe;for(Qe=0,et=1;et<Se.length;)!Se[Qe].endsWith(" ")&&fe.includes(Se[et])?(Se[Qe]+=Se[et],Ge[Qe]=(0,o.mergeArrays)(Ge[Qe],Ge[et]),Xe[Qe]=(0,o.mergeArrays)(Xe[Qe],Xe[et]),Se[et]="",Ge[et]=[],Xe[et]=[]):Qe=et,++et;return[Se.filter(bt=>bt),Ge.filter(bt=>bt.length>0),Xe.filter(bt=>bt.length>0)]}}class fs extends ft{}class Gs extends ft{}class Gr extends ft{}class Ne extends ft{constructor(k,G){super(k,G),this.languageRegex=/^(>>\w+<<)\s*/g,this.supported_language_codes=this.model.vocab.filter(se=>this.languageRegex.test(se)),console.warn('WARNING: `MarianTokenizer` is not yet supported by Hugging Face\'s "fast" tokenizers library. Therefore, you may experience slightly inaccurate results.')}_encode_text(k){if(k===null)return null;const[G,...se]=k.trim().split(this.languageRegex);if(se.length===0)return super._encode_text(G);if(se.length===2){const[ue,fe]=se;return this.supported_language_codes.includes(ue)||console.warn(`Unsupported language code "${ue}" detected, which may lead to unexpected behavior. Should be one of: ${JSON.stringify(this.supported_language_codes)}`),(0,o.mergeArrays)([ue],super._encode_text(fe))}}}class je extends ft{}class rt extends ft{}class Qt extends ft{}class Hs extends ft{}class Ss extends ft{}class ss extends ft{constructor(k,G){super(k,G),this.decoder=new ht({})}}class Tn extends ft{}class En extends ft{}class Pn extends ft{}class Cn{static async from_pretrained(k,{progress_callback:G=null,config:se=null,cache_dir:ue=null,local_files_only:fe=!1,revision:Se="main",legacy:Ge=null}={}){var Rt;const[Xe,Qe]=await d(k,{progress_callback:G,config:se,cache_dir:ue,local_files_only:fe,revision:Se,legacy:Ge}),et=((Rt=Qe.tokenizer_class)==null?void 0:Rt.replace(/Fast$/,""))??"PreTrainedTokenizer";let bt=this.TOKENIZER_CLASS_MAPPING[et];return bt||(console.warn(`Unknown tokenizer class "${et}", attempting to construct from base class.`),bt=ft),new bt(Xe,Qe)}}re(Cn,"TOKENIZER_CLASS_MAPPING",{T5Tokenizer:pt,DistilBertTokenizer:xe,CamembertTokenizer:De,DebertaTokenizer:$,DebertaV2Tokenizer:ee,BertTokenizer:Os,HerbertTokenizer:V,ConvBertTokenizer:Y,RoFormerTokenizer:oe,XLMTokenizer:nt,ElectraTokenizer:wt,MobileBertTokenizer:St,SqueezeBertTokenizer:Kt,AlbertTokenizer:Ds,GPT2Tokenizer:xt,BartTokenizer:tt,MBartTokenizer:It,MBart50Tokenizer:qt,RobertaTokenizer:Wr,WhisperTokenizer:ar,CodeGenTokenizer:fs,CLIPTokenizer:Gs,SiglipTokenizer:Gr,MarianTokenizer:Ne,BloomTokenizer:qr,NllbTokenizer:Cs,M2M100Tokenizer:mr,LlamaTokenizer:kr,CodeLlamaTokenizer:cr,XLMRobertaTokenizer:ps,MPNetTokenizer:hs,FalconTokenizer:Ir,GPTNeoXTokenizer:Ls,EsmTokenizer:zs,Wav2Vec2CTCTokenizer:je,BlenderbotTokenizer:rt,BlenderbotSmallTokenizer:Qt,SpeechT5Tokenizer:Hs,NougatTokenizer:Ss,VitsTokenizer:ss,Qwen2Tokenizer:vr,GemmaTokenizer:ms,Grok1Tokenizer:Yr,CohereTokenizer:Tn,MgpstrTokenizer:En,Ernie4_5_Tokenizer:Pn,PreTrainedTokenizer:ft})},"./src/utils/audio.js":(e,r,t)=>{t.r(r),t.d(r,{RawAudio:()=>U,hamming:()=>u,hanning:()=>d,mel_filter_bank:()=>C,read_audio:()=>c,spectrogram:()=>P,window_function:()=>F});var s=t("./src/utils/hub.js"),o=t("./src/utils/maths.js"),n=t("./src/utils/core.js"),a=t("./src/env.js"),i=t("./src/utils/tensor.js"),l=t("?7992");async function c(j,ne){if(typeof AudioContext>"u")throw Error("Unable to load audio from path/URL since `AudioContext` is not available in your environment. Instead, audio data should be passed directly to the pipeline/processor. For more information and some example code, see https://huggingface.co/docs/transformers.js/guides/node-audio-processing.");const q=await(await(0,s.getFile)(j)).arrayBuffer(),te=new AudioContext({sampleRate:ne});typeof ne>"u"&&console.warn(`No sampling rate provided, using default of ${te.sampleRate}Hz.`);const Z=await te.decodeAudioData(q);let ae;if(Z.numberOfChannels===2){const he=Math.sqrt(2),Q=Z.getChannelData(0),B=Z.getChannelData(1);ae=new Float32Array(Q.length);for(let O=0;O<Z.length;++O)ae[O]=he*(Q[O]+B[O])/2}else ae=Z.getChannelData(0);return ae}function p(j,ne){if(j<1)return new Float64Array;if(j===1)return new Float64Array([1]);const q=1-ne,te=2*Math.PI/(j-1),Z=new Float64Array(j);for(let ae=0;ae<j;++ae)Z[ae]=ne-q*Math.cos(ae*te);return Z}function d(j){return p(j,.5)}function u(j){return p(j,.54)}const _={htk:j=>2595*Math.log10(1+j/700),kaldi:j=>1127*Math.log(1+j/700),slaney:(j,ne=1e3,q=15,te=27/Math.log(6.4))=>j>=ne?q+Math.log(j/ne)*te:3*j/200};function f(j,ne="htk"){const q=_[ne];if(!q)throw new Error('mel_scale should be one of "htk", "slaney" or "kaldi".');return typeof j=="number"?q(j):j.map(te=>q(te))}const b={htk:j=>700*(10**(j/2595)-1),kaldi:j=>700*(Math.exp(j/1127)-1),slaney:(j,ne=1e3,q=15,te=Math.log(6.4)/27)=>j>=q?ne*Math.exp(te*(j-q)):200*j/3};function A(j,ne="htk"){const q=b[ne];if(!q)throw new Error('mel_scale should be one of "htk", "slaney" or "kaldi".');return typeof j=="number"?q(j):j.map(te=>q(te))}function g(j,ne){const q=Float64Array.from({length:ne.length-1},(he,Q)=>ne[Q+1]-ne[Q]),te=Array.from({length:j.length},()=>new Array(ne.length));for(let he=0;he<j.length;++he){const Q=te[he];for(let B=0;B<ne.length;++B)Q[B]=ne[B]-j[he]}const Z=ne.length-2,ae=Array.from({length:Z},()=>new Array(j.length));for(let he=0;he<j.length;++he){const Q=te[he];for(let B=0;B<Z;++B){const O=-Q[B]/q[B],W=Q[B+2]/q[B+1];ae[B][he]=Math.max(0,Math.min(O,W))}}return ae}function y(j,ne,q){const te=(ne-j)/(q-1);return Float64Array.from({length:q},(Z,ae)=>j+te*ae)}function C(j,ne,q,te,Z,ae=null,he="htk",Q=!1){if(ae!==null&&ae!=="slaney")throw new Error('norm must be one of null or "slaney"');if(j<2)throw new Error(`Require num_frequency_bins: ${j} >= 2`);if(q>te)throw new Error(`Require min_frequency: ${q} <= max_frequency: ${te}`);const B=f(q,he),O=f(te,he),W=y(B,O,ne+2);let N=A(W,he),J;if(Q){const me=Z/((j-1)*2);J=f(Float64Array.from({length:j},(Ae,Ve)=>Ve*me),he),N=W}else J=y(0,Math.floor(Z/2),j);const ie=g(J,N);if(ae!==null&&ae==="slaney")for(let me=0;me<ne;++me){const Ae=ie[me],Ve=2/(N[me+2]-N[me]);for(let $e=0;$e<j;++$e)Ae[$e]*=Ve}return ie}function x(j,ne,q){const te=new j.constructor(j.length+ne+q),Z=j.length-1;for(let ae=0;ae<j.length;++ae)te[ne+ae]=j[ae];for(let ae=1;ae<=ne;++ae)te[ne-ae]=j[(0,n.calculateReflectOffset)(ae,Z)];for(let ae=1;ae<=q;++ae)te[Z+ne+ae]=j[(0,n.calculateReflectOffset)(Z-ae,Z)];return te}function M(j,ne,q,te,Z){if(q<=0)throw new Error("reference must be greater than zero");if(te<=0)throw new Error("min_value must be greater than zero");q=Math.max(te,q);const ae=Math.log10(q);for(let he=0;he<j.length;++he)j[he]=ne*Math.log10(Math.max(te,j[he])-ae);if(Z!==null){if(Z<=0)throw new Error("db_range must be greater than zero");const he=(0,o.max)(j)[0]-Z;for(let Q=0;Q<j.length;++Q)j[Q]=Math.max(j[Q],he)}return j}function T(j,ne=1,q=1e-5,te=null){return M(j,20,ne,q,te)}function v(j,ne=1,q=1e-10,te=null){return M(j,10,ne,q,te)}async function P(j,ne,q,te,{fft_length:Z=null,power:ae=1,center:he=!0,pad_mode:Q="reflect",onesided:B=!0,preemphasis:O=null,preemphasis_htk_flavor:W=!0,mel_filters:N=null,mel_floor:J=1e-10,log_mel:ie=null,reference:me=1,min_value:Ae=1e-10,db_range:Ve=null,remove_dc_offset:$e=null,min_num_frames:X=null,max_num_frames:z=null,do_pad:_e=!0,transpose:Ee=!1,mel_offset:Me=0}={}){const Ce=ne.length;if(Z===null&&(Z=q),q>Z)throw Error(`frame_length (${q}) may not be larger than fft_length (${Z})`);if(Ce!==q)throw new Error(`Length of the window (${Ce}) must equal frame_length (${q})`);if(te<=0)throw new Error("hop_length must be greater than zero");if(ae===null&&N!==null)throw new Error("You have provided `mel_filters` but `power` is `None`. Mel spectrogram computation is not yet supported for complex-valued spectrogram. Specify `power` to fix this issue.");if(!W)throw new Error("`preemphasis_htk_flavor=false` is not currently supported.");if(he)switch(Q){case"reflect":{const ze=Math.floor((Z-1)/2)+1;j=x(j,ze,ze);break}case"constant":{const ze=Math.floor(Z/2),He=new j.constructor(j.length+2*ze);He.set(j,ze),j=He;break}default:throw new Error(`pad_mode="${Q}" not implemented yet.`)}let ye=Math.floor(1+Math.floor((j.length-q)/te));X!==null&&ye<X&&(ye=X);const de=B?Math.floor(Z/2)+1:Z;let we=ye,ce=ye;z!==null&&(z>ye?_e&&(ce=z):ce=we=z);const ke=new o.FFT(Z),Le=new Float64Array(Z),Te=new Float64Array(ke.outputBufferSize),We=new Float32Array(de*ce);for(let ze=0;ze<we;++ze){const He=ze*te,gt=Math.min(j.length-He,q);gt!==q&&Le.fill(0,0,q);for(let dt=0;dt<gt;++dt)Le[dt]=j[He+dt];if($e){let dt=0;for(let ht=0;ht<gt;++ht)dt+=Le[ht];const kt=dt/gt;for(let ht=0;ht<gt;++ht)Le[ht]-=kt}if(O!==null){for(let dt=gt-1;dt>=1;--dt)Le[dt]-=O*Le[dt-1];Le[0]*=1-O}for(let dt=0;dt<ne.length;++dt)Le[dt]*=ne[dt];ke.realTransform(Te,Le);for(let dt=0;dt<de;++dt){const kt=dt<<1;We[dt*ce+ze]=Te[kt]**2+Te[kt+1]**2}}if(ae!==null&&ae!==2){const ze=ae/2;for(let He=0;He<We.length;++He)We[He]**=ze}const qe=N.length;let st=await(0,i.matmul)(new i.Tensor("float32",N.flat(),[qe,de]),new i.Tensor("float32",We,[de,ce]));Ee&&(st=st.transpose(1,0));const Ze=st.data;for(let ze=0;ze<Ze.length;++ze)Ze[ze]=Me+Math.max(J,Ze[ze]);if(ae!==null&&ie!==null){const ze=Math.min(Ze.length,we*qe);switch(ie){case"log":for(let He=0;He<ze;++He)Ze[He]=Math.log(Ze[He]);break;case"log10":for(let He=0;He<ze;++He)Ze[He]=Math.log10(Ze[He]);break;case"dB":if(ae===1)T(Ze,me,Ae,Ve);else if(ae===2)v(Ze,me,Ae,Ve);else throw new Error(`Cannot use log_mel option '${ie}' with power ${ae}`);break;default:throw new Error(`log_mel must be one of null, 'log', 'log10' or 'dB'. Got '${ie}'`)}}return st}function F(j,ne,{periodic:q=!0,frame_length:te=null,center:Z=!0}={}){const ae=q?j+1:j;let he;switch(ne){case"boxcar":he=new Float64Array(ae).fill(1);break;case"hann":case"hann_window":he=d(ae);break;case"hamming":he=u(ae);break;case"povey":he=d(ae).map(Q=>Math.pow(Q,.85));break;default:throw new Error(`Unknown window type ${ne}.`)}if(q&&(he=he.subarray(0,j)),te===null)return he;if(j>te)throw new Error(`Length of the window (${j}) may not be larger than frame_length (${te})`);return he}function D(j,ne){let q=44;const te=new ArrayBuffer(q+j.length*4),Z=new DataView(te);K(Z,0,"RIFF"),Z.setUint32(4,36+j.length*4,!0),K(Z,8,"WAVE"),K(Z,12,"fmt "),Z.setUint32(16,16,!0),Z.setUint16(20,3,!0),Z.setUint16(22,1,!0),Z.setUint32(24,ne,!0),Z.setUint32(28,ne*4,!0),Z.setUint16(32,4,!0),Z.setUint16(34,32,!0),K(Z,36,"data"),Z.setUint32(40,j.length*4,!0);for(let ae=0;ae<j.length;++ae,q+=4)Z.setFloat32(q,j[ae],!0);return te}function K(j,ne,q){for(let te=0;te<q.length;++te)j.setUint8(ne+te,q.charCodeAt(te))}class U{constructor(ne,q){this.audio=ne,this.sampling_rate=q}toWav(){return D(this.audio,this.sampling_rate)}toBlob(){const ne=this.toWav();return new Blob([ne],{type:"audio/wav"})}async save(ne){let q;if(a.apis.IS_BROWSER_ENV){if(a.apis.IS_WEBWORKER_ENV)throw new Error("Unable to save a file from a Web Worker.");q=n.saveBlob}else if(a.apis.IS_FS_AVAILABLE)q=async(te,Z)=>{let ae=await Z.arrayBuffer();l.writeFileSync(te,Buffer.from(ae))};else throw new Error("Unable to save because filesystem is disabled in this environment.");await q(ne,this.toBlob())}}},"./src/utils/constants.js":(e,r,t)=>{t.r(r),t.d(r,{CHAT_TEMPLATE_NAME:()=>l,CONFIG_NAME:()=>o,FEATURE_EXTRACTOR_NAME:()=>n,GENERATION_CONFIG_NAME:()=>c,GITHUB_ISSUE_URL:()=>s,IMAGE_PROCESSOR_NAME:()=>a,PROCESSOR_NAME:()=>i});const s="https://github.com/huggingface/transformers.js/issues/new/choose",o="config.json",n="preprocessor_config.json",a=n,i="processor_config.json",l="chat_template.jinja",c="generation_config.json"},"./src/utils/core.js":(e,r,t)=>{t.r(r),t.d(r,{calculateDimensions:()=>c,calculateReflectOffset:()=>_,count:()=>g,dispatchCallback:()=>s,escapeRegExp:()=>n,isIntegralNumber:()=>i,isNullishDimension:()=>l,isTypedArray:()=>a,len:()=>A,mergeArrays:()=>d,pick:()=>b,pop:()=>p,product:()=>u,reverseDictionary:()=>o,saveBlob:()=>f});function s(y,C){y&&y(C)}function o(y){return Object.fromEntries(Object.entries(y).map(([C,x])=>[x,C]))}function n(y){return y.replace(/[.*+?^${}()|[\]\\]/g,"\\$&")}function a(y){var C,x,M;return((M=(x=(C=y==null?void 0:y.prototype)==null?void 0:C.__proto__)==null?void 0:x.constructor)==null?void 0:M.name)==="TypedArray"}function i(y){return Number.isInteger(y)||typeof y=="bigint"}function l(y){return y==null||y===-1}function c(y){const C=[];let x=y;for(;Array.isArray(x);)C.push(x.length),x=x[0];return C}function p(y,C,x=void 0){const M=y[C];if(M!==void 0)return delete y[C],M;if(x===void 0)throw Error(`Key ${C} does not exist in object.`);return x}function d(...y){return Array.prototype.concat.apply([],y)}function u(...y){return y.reduce((C,x)=>C.flatMap(M=>x.map(T=>[M,T])))}function _(y,C){return Math.abs((y+C)%(2*C)-C)}function f(y,C){const x=URL.createObjectURL(C),M=document.createElement("a");M.href=x,M.download=y,M.click(),M.remove(),URL.revokeObjectURL(x)}function b(y,C){return Object.assign({},...C.map(x=>{if(y[x]!==void 0)return{[x]:y[x]}}))}function A(y){let C=0;for(const x of y)++C;return C}function g(y,C){let x=0;for(const M of y)M===C&&++x;return x}},"./src/utils/data-structures.js":(e,r,t)=>{t.r(r),t.d(r,{CharTrie:()=>o,DictionarySplitter:()=>l,LRUCache:()=>c,PriorityQueue:()=>s,TokenLattice:()=>a});class s{constructor(d=(_,f)=>_>f,u=1/0){this._heap=[],this._comparator=d,this._maxSize=u}get size(){return this._heap.length}isEmpty(){return this.size===0}peek(){return this._heap[0]}push(...d){return this.extend(d)}extend(d){for(const u of d)if(this.size<this._maxSize)this._heap.push(u),this._siftUp();else{const _=this._smallest();this._comparator(u,this._heap[_])&&(this._heap[_]=u,this._siftUpFrom(_))}return this.size}pop(){const d=this.peek(),u=this.size-1;return u>0&&this._swap(0,u),this._heap.pop(),this._siftDown(),d}replace(d){const u=this.peek();return this._heap[0]=d,this._siftDown(),u}_parent(d){return(d+1>>>1)-1}_left(d){return(d<<1)+1}_right(d){return d+1<<1}_greater(d,u){return this._comparator(this._heap[d],this._heap[u])}_swap(d,u){const _=this._heap[d];this._heap[d]=this._heap[u],this._heap[u]=_}_siftUp(){this._siftUpFrom(this.size-1)}_siftUpFrom(d){for(;d>0&&this._greater(d,this._parent(d));)this._swap(d,this._parent(d)),d=this._parent(d)}_siftDown(){let d=0;for(;this._left(d)<this.size&&this._greater(this._left(d),d)||this._right(d)<this.size&&this._greater(this._right(d),d);){const u=this._right(d)<this.size&&this._greater(this._right(d),this._left(d))?this._right(d):this._left(d);this._swap(d,u),d=u}}_smallest(){return 2**Math.floor(Math.log2(this.size))-1}}class o{constructor(){this.root=n.default()}extend(d){for(const u of d)this.push(u)}push(d){let u=this.root;for(const _ of d){let f=u.children.get(_);f===void 0&&(f=n.default(),u.children.set(_,f)),u=f}u.isLeaf=!0}*commonPrefixSearch(d){let u=this.root;if(u===void 0)return;let _="";for(const f of d){if(_+=f,u=u.children.get(f),u===void 0)return;u.isLeaf&&(yield _)}}}class n{constructor(d,u){this.isLeaf=d,this.children=u}static default(){return new n(!1,new Map)}}class a{constructor(d,u,_){this.chars=Array.from(d),this.len=this.chars.length,this.bosTokenId=u,this.eosTokenId=_,this.nodes=[],this.beginNodes=Array.from({length:this.len+1},()=>[]),this.endNodes=Array.from({length:this.len+1},()=>[]);const f=new i(this.bosTokenId,0,0,0,0),b=new i(this.eosTokenId,1,this.len,0,0);this.nodes.push(f.clone()),this.nodes.push(b.clone()),this.beginNodes[this.len].push(b),this.endNodes[0].push(f)}insert(d,u,_,f){const b=this.nodes.length,A=new i(f,b,d,u,_);this.beginNodes[d].push(A),this.endNodes[d+u].push(A),this.nodes.push(A)}viterbi(){const d=this.len;let u=0;for(;u<=d;){if(this.beginNodes[u].length==0)return[];for(let g of this.beginNodes[u]){g.prev=null;let y=0,C=null;for(let x of this.endNodes[u]){const M=x.backtraceScore+g.score;(C===null||M>y)&&(C=x.clone(),y=M)}if(C!==null)g.prev=C,g.backtraceScore=y;else return[]}++u}const _=[],b=this.beginNodes[d][0].prev;if(b===null)return[];let A=b.clone();for(;A.prev!==null;)_.push(A.clone()),A=A.clone().prev.clone();return _.reverse(),_}piece(d){return this.chars.slice(d.pos,d.pos+d.length).join("")}tokens(){return this.viterbi().map(u=>this.piece(u))}tokenIds(){return this.viterbi().map(u=>u.tokenId)}}class i{constructor(d,u,_,f,b){this.tokenId=d,this.nodeId=u,this.pos=_,this.length=f,this.score=b,this.prev=null,this.backtraceScore=0}clone(){const d=new i(this.tokenId,this.nodeId,this.pos,this.length,this.score);return d.prev=this.prev,d.backtraceScore=this.backtraceScore,d}}class l{constructor(d){this.trie=this._buildTrie(d)}_buildTrie(d){var _;const u=Object.create(null);for(const f of d){let b=u;for(let A=0;A<f.length;++A)b=b[_=f[A]]??(b[_]=Object.create(null));b.end=f}return u}split(d){const u=[],_=d.length;let f=0,b=0;for(;b<_;){let A=this.trie,g=null,y=b;for(;y<_&&(A=A[d[y]]);)A.end&&(g=A.end),++y;g?(b>f&&u.push(d.slice(f,b)),u.push(g),b+=g.length,f=b):++b}return f<_&&u.push(d.slice(f)),u}}class c{constructor(d){this.capacity=d,this.cache=new Map}get(d){if(!this.cache.has(d))return;const u=this.cache.get(d);return this.cache.delete(d),this.cache.set(d,u),u}put(d,u){this.cache.has(d)&&this.cache.delete(d),this.cache.set(d,u),this.cache.size>this.capacity&&this.cache.delete(this.cache.keys().next().value)}clear(){this.cache.clear()}}},"./src/utils/devices.js":(e,r,t)=>{t.r(r),t.d(r,{DEVICE_TYPES:()=>s});const s=Object.freeze({auto:"auto",gpu:"gpu",cpu:"cpu",wasm:"wasm",webgpu:"webgpu",cuda:"cuda",dml:"dml",webnn:"webnn","webnn-npu":"webnn-npu","webnn-gpu":"webnn-gpu","webnn-cpu":"webnn-cpu"})},"./src/utils/dtypes.js":(e,r,t)=>{t.r(r),t.d(r,{DATA_TYPES:()=>a,DEFAULT_DEVICE_DTYPE_MAPPING:()=>i,DEFAULT_DTYPE_SUFFIX_MAPPING:()=>l,isWebGpuFp16Supported:()=>n});var s=t("./src/env.js"),o=t("./src/utils/devices.js");const n=function(){let c;return async function(){if(c===void 0)if(!s.apis.IS_WEBGPU_AVAILABLE)c=!1;else try{c=(await navigator.gpu.requestAdapter()).features.has("shader-f16")}catch{c=!1}return c}}(),a=Object.freeze({auto:"auto",fp32:"fp32",fp16:"fp16",q8:"q8",int8:"int8",uint8:"uint8",q4:"q4",bnb4:"bnb4",q4f16:"q4f16"}),i=Object.freeze({[o.DEVICE_TYPES.wasm]:a.q8}),l=Object.freeze({[a.fp32]:"",[a.fp16]:"_fp16",[a.int8]:"_int8",[a.uint8]:"_uint8",[a.q8]:"_quantized",[a.q4]:"_q4",[a.q4f16]:"_q4f16",[a.bnb4]:"_bnb4"})},"./src/utils/generic.js":(e,r,t)=>{t.r(r),t.d(r,{Callable:()=>s});const s=class{constructor(){let o=function(...n){return o._call(...n)};return Object.setPrototypeOf(o,new.target.prototype)}_call(...o){throw Error("Must implement _call method in subclass")}}},"./src/utils/hub.js":(e,r,t)=>{t.r(r),t.d(r,{MAX_EXTERNAL_DATA_CHUNKS:()=>i,getFile:()=>_,getModelFile:()=>y,getModelJSON:()=>x,getModelText:()=>C});var s=t("?7992"),o=t("?5af5"),n=t("./src/env.js"),a=t("./src/utils/core.js");const i=100,l={txt:"text/plain",html:"text/html",css:"text/css",js:"text/javascript",json:"application/json",png:"image/png",jpg:"image/jpeg",jpeg:"image/jpeg",gif:"image/gif"};class c{constructor(P){if(this.filePath=P,this.headers=new Headers,this.exists=s.existsSync(P),this.exists){this.status=200,this.statusText="OK";let F=s.statSync(P);this.headers.set("content-length",F.size.toString()),this.updateContentType();const D=s.createReadStream(P);this.body=new ReadableStream({start(K){D.on("data",U=>K.enqueue(U)),D.on("end",()=>K.close()),D.on("error",U=>K.error(U))},cancel(){D.destroy()}})}else this.status=404,this.statusText="Not Found",this.body=null}updateContentType(){const P=this.filePath.toString().split(".").pop().toLowerCase();this.headers.set("content-type",l[P]??"application/octet-stream")}clone(){let P=new c(this.filePath);return P.exists=this.exists,P.status=this.status,P.statusText=this.statusText,P.headers=new Headers(this.headers),P}async arrayBuffer(){return(await s.promises.readFile(this.filePath)).buffer}async blob(){const P=await s.promises.readFile(this.filePath);return new Blob([P],{type:this.headers.get("content-type")})}async text(){return await s.promises.readFile(this.filePath,"utf8")}async json(){return JSON.parse(await this.text())}}function p(v,P=null,F=null){let D;try{D=new URL(v)}catch{return!1}return!(P&&!P.includes(D.protocol)||F&&!F.includes(D.hostname))}const d=/^(\b[\w\-.]+\b\/)?\b[\w\-.]{1,96}\b$/;function u(v){return!(!d.test(v)||v.includes("..")||v.includes("--")||v.endsWith(".git")||v.endsWith(".ipynb"))}async function _(v){var P;if(n.env.useFS&&!p(v,["http:","https:","blob:"]))return new c(v instanceof URL?v.protocol==="file:"?v.pathname:v.toString():v);if(typeof process<"u"&&((P=process==null?void 0:process.release)==null?void 0:P.name)==="node"){const F=!!(Us!=null&&Us.TESTING_REMOTELY),D=n.env.version,K=new Headers;if(K.set("User-Agent",`transformers.js/${D}; is_ci/${F};`),p(v,["http:","https:"],["huggingface.co","hf.co"])){const j=(Us==null?void 0:Us.HF_TOKEN)??(Us==null?void 0:Us.HF_ACCESS_TOKEN);j&&K.set("Authorization",`Bearer ${j}`)}return fetch(v,{headers:K})}else return fetch(v)}const f={400:"Bad request error occurred while trying to load file",401:"Unauthorized access to file",403:"Forbidden access to file",404:"Could not locate file",408:"Request timeout error occurred while trying to load file",500:"Internal server error error occurred while trying to load file",502:"Bad gateway error occurred while trying to load file",503:"Service unavailable error occurred while trying to load file",504:"Gateway timeout error occurred while trying to load file"};function b(v,P,F){if(!F)return null;const D=f[v]??`Error (${v}) occurred while trying to load file`;throw Error(`${D}: "${P}".`)}class A{constructor(P){this.path=P}async match(P){let F=o.join(this.path,P),D=new c(F);if(D.exists)return D}async put(P,F,D=void 0){let K=o.join(this.path,P);try{const U=F.headers.get("Content-Length"),j=parseInt(U??"0");let ne=0;await s.promises.mkdir(o.dirname(K),{recursive:!0});const q=s.createWriteStream(K),te=F.body.getReader();for(;;){const{done:Z,value:ae}=await te.read();if(Z)break;await new Promise((Q,B)=>{q.write(ae,O=>{if(O){B(O);return}Q()})}),ne+=ae.length;const he=j?ne/j*100:0;D==null||D({progress:he,loaded:ne,total:j})}q.close()}catch(U){try{await s.promises.unlink(K)}catch{}throw U}}}async function g(v,...P){for(let F of P)try{let D=await v.match(F);if(D)return D}catch{continue}}async function y(v,P,F=!0,D={},K=!1){if(!n.env.allowLocalModels){if(D.local_files_only)throw Error("Invalid configuration detected: local models are disabled (`env.allowLocalModels=false`) but you have requested to only use local models (`local_files_only=true`).");if(!n.env.allowRemoteModels)throw Error("Invalid configuration detected: both local and remote models are disabled. Fix by setting `env.allowLocalModels` or `env.allowRemoteModels` to `true`.")}(0,a.dispatchCallback)(D.progress_callback,{status:"initiate",name:v,file:P});let U;if(!U&&n.env.useCustomCache){if(!n.env.customCache)throw Error("`env.useCustomCache=true`, but `env.customCache` is not defined.");if(!n.env.customCache.match||!n.env.customCache.put)throw new Error("`env.customCache` must be an object which implements the `match` and `put` functions of the Web Cache API. For more information, see https://developer.mozilla.org/en-US/docs/Web/API/Cache");U=n.env.customCache}if(!U&&n.env.useBrowserCache){if(typeof caches>"u")throw Error("Browser cache is not available in this environment.");try{U=await caches.open("transformers-cache")}catch(J){console.warn("An error occurred while opening the browser cache:",J)}}if(!U&&n.env.useFSCache){if(!n.apis.IS_FS_AVAILABLE)throw Error("File System Cache is not available in this environment.");U=new A(D.cache_dir??n.env.cacheDir)}const j=D.revision??"main",ne=T(v,P),q=u(v),te=q?T(n.env.localModelPath,ne):ne,Z=T(n.env.remoteHost,n.env.remotePathTemplate.replaceAll("{model}",v).replaceAll("{revision}",encodeURIComponent(j)),P);let ae;const he=U instanceof A?j==="main"?ne:T(v,j,P):Z;let Q=!1,B;U&&(B=await g(U,te,he));const O=B!==void 0;if(B===void 0){if(n.env.allowLocalModels)if(p(ne,["http:","https:"])){if(D.local_files_only)throw new Error(`\`local_files_only=true\`, but attempted to load a remote file from: ${ne}.`);if(!n.env.allowRemoteModels)throw new Error(`\`env.allowRemoteModels=false\`, but attempted to load a remote file from: ${ne}.`)}else try{B=await _(te),ae=te}catch(ie){console.warn(`Unable to load from local path "${te}": "${ie}"`)}if(B===void 0||B.status===404){if(D.local_files_only||!n.env.allowRemoteModels){if(F)throw Error(`\`local_files_only=true\` or \`env.allowRemoteModels=false\` and file was not found locally at "${te}".`);return null}if(!q)throw Error(`Local file missing at "${te}" and download aborted due to invalid model ID "${v}".`);if(B=await _(Z),B.status!==200)return b(B.status,Z,F);ae=he}Q=U&&typeof Response<"u"&&B instanceof Response&&B.status===200}(0,a.dispatchCallback)(D.progress_callback,{status:"download",name:v,file:P});let W;if(!(n.apis.IS_NODE_ENV&&K)){let J;D.progress_callback?O&&typeof navigator<"u"&&/firefox/i.test(navigator.userAgent)?(J=new Uint8Array(await B.arrayBuffer()),(0,a.dispatchCallback)(D.progress_callback,{status:"progress",name:v,file:P,progress:100,loaded:J.length,total:J.length})):J=await M(B,ie=>{(0,a.dispatchCallback)(D.progress_callback,{status:"progress",name:v,file:P,...ie})}):J=new Uint8Array(await B.arrayBuffer()),W=J}if(Q&&ae&&await U.match(ae)===void 0)if(W)await U.put(ae,new Response(W,{headers:B.headers})).catch(J=>{console.warn(`Unable to add response to browser cache: ${J}.`)});else{const J=D.progress_callback?ie=>(0,a.dispatchCallback)(D.progress_callback,{status:"progress",name:v,file:P,...ie}):void 0;await U.put(ae,B,J)}if((0,a.dispatchCallback)(D.progress_callback,{status:"done",name:v,file:P}),W){if(!n.apis.IS_NODE_ENV&&K)throw new Error("Cannot return path in a browser environment.");return W}if(B instanceof c)return B.filePath;const N=await(U==null?void 0:U.match(ae));if(N instanceof c)return N.filePath;if(N instanceof Response)return new Uint8Array(await N.arrayBuffer());if(typeof N=="string")return N;throw new Error("Unable to get model file path or buffer.")}async function C(v,P,F=!0,D={}){const K=await y(v,P,F,D,!1);return K===null?null:new TextDecoder("utf-8").decode(K)}async function x(v,P,F=!0,D={}){const K=await C(v,P,F,D);return K===null?{}:JSON.parse(K)}async function M(v,P){const F=v.headers.get("Content-Length");F===null&&console.warn("Unable to determine content-length from response headers. Will expand buffer when needed.");let D=parseInt(F??"0"),K=new Uint8Array(D),U=0;const j=v.body.getReader();async function ne(){const{done:q,value:te}=await j.read();if(q)return;const Z=U+te.length;if(Z>D){D=Z;const he=new Uint8Array(D);he.set(K),K=he}K.set(te,U),U=Z;const ae=U/D*100;return P({progress:ae,loaded:U,total:D}),ne()}return await ne(),K}function T(...v){return v=v.map((P,F)=>(F&&(P=P.replace(new RegExp("^/"),"")),F!==v.length-1&&(P=P.replace(new RegExp("/$"),"")),P)),v.join("/")}},"./src/utils/image.js":(e,r,t)=>{t.r(r),t.d(r,{RawImage:()=>f,load_image:()=>b});var s=t("./src/utils/core.js"),o=t("./src/utils/hub.js"),n=t("./src/env.js"),a=t("./src/utils/tensor.js"),i=t("?2b25");let l,c,p;const d=n.apis.IS_BROWSER_ENV||n.apis.IS_WEBWORKER_ENV;if(d)l=(A,g)=>{if(!self.OffscreenCanvas)throw new Error("OffscreenCanvas not supported by this browser.");return new self.OffscreenCanvas(A,g)},p=self.createImageBitmap,c=self.ImageData;else if(i)p=async A=>{const y=(await A.metadata()).channels,{data:C,info:x}=await A.rotate().raw().toBuffer({resolveWithObject:!0}),M=new f(new Uint8ClampedArray(C),x.width,x.height,x.channels);return y!==void 0&&y!==x.channels&&M.convert(y),M};else throw new Error("Unable to load image processing library.");const u={0:"nearest",1:"lanczos",2:"bilinear",3:"bicubic",4:"box",5:"hamming"},_=new Map([["png","image/png"],["jpg","image/jpeg"],["jpeg","image/jpeg"],["gif","image/gif"]]);class f{constructor(g,y,C,x){this.data=g,this.width=y,this.height=C,this.channels=x}get size(){return[this.width,this.height]}static async read(g){if(g instanceof f)return g;if(typeof g=="string"||g instanceof URL)return await this.fromURL(g);if(g instanceof Blob)return await this.fromBlob(g);if(typeof HTMLCanvasElement<"u"&&g instanceof HTMLCanvasElement||typeof OffscreenCanvas<"u"&&g instanceof OffscreenCanvas)return this.fromCanvas(g);throw new Error(`Unsupported input type: ${typeof g}`)}static fromCanvas(g){if(!d)throw new Error("fromCanvas() is only supported in browser environments.");const C=g.getContext("2d").getImageData(0,0,g.width,g.height).data;return new f(C,g.width,g.height,4)}static async fromURL(g){const y=await(0,o.getFile)(g);if(y.status!==200)throw new Error(`Unable to read image from "${g}" (${y.status} ${y.statusText})`);const C=await y.blob();return this.fromBlob(C)}static async fromBlob(g){if(d){const y=await p(g),C=l(y.width,y.height).getContext("2d");return C.drawImage(y,0,0),new this(C.getImageData(0,0,y.width,y.height).data,y.width,y.height,4)}else{const y=i(await g.arrayBuffer());return await p(y)}}static fromTensor(g,y="CHW"){if(g.dims.length!==3)throw new Error(`Tensor should have 3 dimensions, but has ${g.dims.length} dimensions.`);if(y==="CHW")g=g.transpose(1,2,0);else if(y!=="HWC")throw new Error(`Unsupported channel format: ${y}`);if(!(g.data instanceof Uint8ClampedArray||g.data instanceof Uint8Array))throw new Error(`Unsupported tensor type: ${g.type}`);switch(g.dims[2]){case 1:case 2:case 3:case 4:return new f(g.data,g.dims[1],g.dims[0],g.dims[2]);default:throw new Error(`Unsupported number of channels: ${g.dims[2]}`)}}grayscale(){if(this.channels===1)return this;const g=new Uint8ClampedArray(this.width*this.height*1);switch(this.channels){case 3:case 4:for(let y=0,C=0;y<this.data.length;y+=this.channels){const x=this.data[y],M=this.data[y+1],T=this.data[y+2];g[C++]=Math.round(.2989*x+.587*M+.114*T)}break;default:throw new Error(`Conversion failed due to unsupported number of channels: ${this.channels}`)}return this._update(g,this.width,this.height,1)}rgb(){if(this.channels===3)return this;const g=new Uint8ClampedArray(this.width*this.height*3);switch(this.channels){case 1:for(let y=0,C=0;y<this.data.length;++y)g[C++]=this.data[y],g[C++]=this.data[y],g[C++]=this.data[y];break;case 4:for(let y=0,C=0;y<this.data.length;y+=4)g[C++]=this.data[y],g[C++]=this.data[y+1],g[C++]=this.data[y+2];break;default:throw new Error(`Conversion failed due to unsupported number of channels: ${this.channels}`)}return this._update(g,this.width,this.height,3)}rgba(){if(this.channels===4)return this;const g=new Uint8ClampedArray(this.width*this.height*4);switch(this.channels){case 1:for(let y=0,C=0;y<this.data.length;++y)g[C++]=this.data[y],g[C++]=this.data[y],g[C++]=this.data[y],g[C++]=255;break;case 3:for(let y=0,C=0;y<this.data.length;y+=3)g[C++]=this.data[y],g[C++]=this.data[y+1],g[C++]=this.data[y+2],g[C++]=255;break;default:throw new Error(`Conversion failed due to unsupported number of channels: ${this.channels}`)}return this._update(g,this.width,this.height,4)}putAlpha(g){if(g.width!==this.width||g.height!==this.height)throw new Error(`Expected mask size to be ${this.width}x${this.height}, but got ${g.width}x${g.height}`);if(g.channels!==1)throw new Error(`Expected mask to have 1 channel, but got ${g.channels}`);const y=this.data,C=g.data,x=this.width*this.height;if(this.channels===3){const M=new Uint8ClampedArray(x*4);for(let T=0,v=0,P=0;T<x;++T)M[P++]=y[v++],M[P++]=y[v++],M[P++]=y[v++],M[P++]=C[T];return this._update(M,this.width,this.height,4)}else if(this.channels===4){for(let M=0;M<x;++M)y[4*M+3]=C[M];return this}throw new Error(`Expected image to have 3 or 4 channels, but got ${this.channels}`)}async resize(g,y,{resample:C=2}={}){if(this.width===g&&this.height===y)return this;let x=u[C]??C;const M=(0,s.isNullishDimension)(g),T=(0,s.isNullishDimension)(y);if(M&&T)return this;if(M?g=y/this.height*this.width:T&&(y=g/this.width*this.height),d){const v=this.channels,P=this.toCanvas(),F=l(g,y).getContext("2d");return F.drawImage(P,0,0,g,y),new f(F.getImageData(0,0,g,y).data,g,y,4).convert(v)}else{let v=this.toSharp();switch(x){case"box":case"hamming":(x==="box"||x==="hamming")&&(console.warn(`Resampling method ${x} is not yet supported. Using bilinear instead.`),x="bilinear");case"nearest":case"bilinear":case"bicubic":v=v.affine([g/this.width,0,0,y/this.height],{interpolator:x});break;case"lanczos":v=v.resize({width:g,height:y,fit:"fill",kernel:"lanczos3"});break;default:throw new Error(`Resampling method ${x} is not supported.`)}return await p(v)}}async pad([g,y,C,x]){if(g=Math.max(g,0),y=Math.max(y,0),C=Math.max(C,0),x=Math.max(x,0),g===0&&y===0&&C===0&&x===0)return this;if(d){const M=this.channels,T=this.toCanvas(),v=this.width+g+y,P=this.height+C+x,F=l(v,P).getContext("2d");return F.drawImage(T,0,0,this.width,this.height,g,C,this.width,this.height),new f(F.getImageData(0,0,v,P).data,v,P,4).convert(M)}else{const M=this.toSharp().extend({left:g,right:y,top:C,bottom:x});return await p(M)}}async crop([g,y,C,x]){if(g=Math.max(g,0),y=Math.max(y,0),C=Math.min(C,this.width-1),x=Math.min(x,this.height-1),g===0&&y===0&&C===this.width-1&&x===this.height-1)return this;const M=C-g+1,T=x-y+1;if(d){const v=this.channels,P=this.toCanvas(),F=l(M,T).getContext("2d");return F.drawImage(P,g,y,M,T,0,0,M,T),new f(F.getImageData(0,0,M,T).data,M,T,4).convert(v)}else{const v=this.toSharp().extract({left:g,top:y,width:M,height:T});return await p(v)}}async center_crop(g,y){if(this.width===g&&this.height===y)return this;const C=(this.width-g)/2,x=(this.height-y)/2;if(d){const M=this.channels,T=this.toCanvas(),v=l(g,y).getContext("2d");let P=0,F=0,D=0,K=0;return C>=0?P=C:D=-C,x>=0?F=x:K=-x,v.drawImage(T,P,F,g,y,D,K,g,y),new f(v.getImageData(0,0,g,y).data,g,y,4).convert(M)}else{let M=this.toSharp();if(C>=0&&x>=0)M=M.extract({left:Math.floor(C),top:Math.floor(x),width:g,height:y});else if(C<=0&&x<=0){const T=Math.floor(-x),v=Math.floor(-C);M=M.extend({top:T,left:v,right:g-this.width-v,bottom:y-this.height-T})}else{let T=[0,0],v=0;x<0?(T[0]=Math.floor(-x),T[1]=y-this.height-T[0]):v=Math.floor(x);let P=[0,0],F=0;C<0?(P[0]=Math.floor(-C),P[1]=g-this.width-P[0]):F=Math.floor(C),M=M.extend({top:T[0],bottom:T[1],left:P[0],right:P[1]}).extract({left:F,top:v,width:g,height:y})}return await p(M)}}async toBlob(g="image/png",y=1){if(!d)throw new Error("toBlob() is only supported in browser environments.");return await this.toCanvas().convertToBlob({type:g,quality:y})}toTensor(g="CHW"){let y=new a.Tensor("uint8",new Uint8Array(this.data),[this.height,this.width,this.channels]);if(g!=="HWC")if(g==="CHW")y=y.permute(2,0,1);else throw new Error(`Unsupported channel format: ${g}`);return y}toCanvas(){if(!d)throw new Error("toCanvas() is only supported in browser environments.");const g=this.clone().rgba(),y=l(g.width,g.height),C=new c(g.data,g.width,g.height);return y.getContext("2d").putImageData(C,0,0),y}split(){const{data:g,width:y,height:C,channels:x}=this,M=g.constructor,T=g.length/x,v=Array.from({length:x},()=>new M(T));for(let P=0;P<T;++P){const F=x*P;for(let D=0;D<x;++D)v[D][P]=g[F+D]}return v.map(P=>new f(P,y,C,1))}_update(g,y,C,x=null){return this.data=g,this.width=y,this.height=C,x!==null&&(this.channels=x),this}clone(){return new f(this.data.slice(),this.width,this.height,this.channels)}convert(g){if(this.channels===g)return this;switch(g){case 1:this.grayscale();break;case 3:this.rgb();break;case 4:this.rgba();break;default:throw new Error(`Conversion failed due to unsupported number of channels: ${this.channels}`)}return this}async save(g){if(d){if(n.apis.IS_WEBWORKER_ENV)throw new Error("Unable to save an image from a Web Worker.");const y=g.split(".").pop().toLowerCase(),C=_.get(y)??"image/png",x=await this.toBlob(C);(0,s.saveBlob)(g,x)}else{if(n.apis.IS_FS_AVAILABLE)return await this.toSharp().toFile(g);throw new Error("Unable to save the image because filesystem is disabled in this environment.")}}toSharp(){if(d)throw new Error("toSharp() is only supported in server-side environments.");return i(this.data,{raw:{width:this.width,height:this.height,channels:this.channels}})}}const b=f.read.bind(f)},"./src/utils/maths.js":(e,r,t)=>{t.r(r),t.d(r,{FFT:()=>b,bankers_round:()=>y,cos_sim:()=>l,dot:()=>i,dynamic_time_warping:()=>C,interpolate_data:()=>s,log_softmax:()=>a,magnitude:()=>c,max:()=>d,medianFilter:()=>A,min:()=>p,permute_data:()=>o,round:()=>g,softmax:()=>n});function s(x,[M,T,v],[P,F],D="bilinear",K=!1){const U=F/v,j=P/T,ne=new x.constructor(P*F*M),q=T*v,te=P*F;for(let Z=0;Z<P;++Z)for(let ae=0;ae<F;++ae){const he=Z*F+ae,Q=(ae+.5)/U-.5,B=(Z+.5)/j-.5;let O=Math.floor(Q),W=Math.floor(B);const N=Math.min(O+1,v-1),J=Math.min(W+1,T-1);O=Math.max(O,0),W=Math.max(W,0);const ie=Q-O,me=B-W,Ae=(1-ie)*(1-me),Ve=ie*(1-me),$e=(1-ie)*me,X=ie*me,z=W*v,_e=J*v,Ee=z+O,Me=z+N,Ce=_e+O,ye=_e+N;for(let de=0;de<M;++de){const we=de*q;ne[de*te+he]=Ae*x[we+Ee]+Ve*x[we+Me]+$e*x[we+Ce]+X*x[we+ye]}}return ne}function o(x,M,T){const v=new Array(T.length),P=new Array(T.length);for(let K=T.length-1,U=1;K>=0;--K)P[K]=U,v[K]=M[T[K]],U*=v[K];const F=T.map((K,U)=>P[T.indexOf(U)]),D=new x.constructor(x.length);for(let K=0;K<x.length;++K){let U=0;for(let j=M.length-1,ne=K;j>=0;--j)U+=ne%M[j]*F[j],ne=Math.floor(ne/M[j]);D[U]=x[K]}return[D,v]}function n(x){const M=d(x)[0],T=x.map(F=>Math.exp(F-M)),v=T.reduce((F,D)=>F+D,0);return T.map(F=>F/v)}function a(x){const M=d(x)[0];let T=0;for(let F=0;F<x.length;++F)T+=Math.exp(x[F]-M);const v=Math.log(T);return x.map(F=>F-M-v)}function i(x,M){let T=0;for(let v=0;v<x.length;++v)T+=x[v]*M[v];return T}function l(x,M){const T=i(x,M),v=c(x),P=c(M);return T/(v*P)}function c(x){return Math.sqrt(x.reduce((M,T)=>M+T*T,0))}function p(x){if(x.length===0)throw Error("Array must not be empty");let M=x[0],T=0;for(let v=1;v<x.length;++v)x[v]<M&&(M=x[v],T=v);return[M,T]}function d(x){if(x.length===0)throw Error("Array must not be empty");let M=x[0],T=0;for(let v=1;v<x.length;++v)x[v]>M&&(M=x[v],T=v);return[M,T]}function u(x){return x>0&&(x&x-1)===0}class _{constructor(M){if(this.size=M|0,this.size<=1||!u(this.size))throw new Error("FFT size must be a power of two larger than 1");this._csize=M<<1,this.table=new Float64Array(this.size*2);for(let v=0;v<this.table.length;v+=2){const P=Math.PI*v/this.size;this.table[v]=Math.cos(P),this.table[v+1]=-Math.sin(P)}let T=0;for(let v=1;this.size>v;v<<=1)++T;this._width=T%2===0?T-1:T,this._bitrev=new Int32Array(1<<this._width);for(let v=0;v<this._bitrev.length;++v){this._bitrev[v]=0;for(let P=0;P<this._width;P+=2){const F=this._width-P-2;this._bitrev[v]|=(v>>>P&3)<<F}}}createComplexArray(){return new Float64Array(this._csize)}fromComplexArray(M,T){const v=T||new Array(M.length>>>1);for(let P=0;P<M.length;P+=2)v[P>>>1]=M[P];return v}toComplexArray(M,T){const v=T||this.createComplexArray();for(let P=0;P<v.length;P+=2)v[P]=M[P>>>1],v[P+1]=0;return v}transform(M,T){if(M===T)throw new Error("Input and output buffers must be different");this._transform4(M,T,1)}realTransform(M,T){if(M===T)throw new Error("Input and output buffers must be different");this._realTransform4(M,T,1)}inverseTransform(M,T){if(M===T)throw new Error("Input and output buffers must be different");this._transform4(M,T,-1);for(let v=0;v<M.length;++v)M[v]/=this.size}_transform4(M,T,v){const P=this._csize;let D=1<<this._width,K=P/D<<1,U,j;const ne=this._bitrev;if(K===4)for(U=0,j=0;U<P;U+=K,++j){const te=ne[j];this._singleTransform2(T,M,U,te,D)}else for(U=0,j=0;U<P;U+=K,++j){const te=ne[j];this._singleTransform4(T,M,U,te,D,v)}const q=this.table;for(D>>=2;D>=2;D>>=2){K=P/D<<1;const te=K>>>2;for(U=0;U<P;U+=K){const Z=U+te-1;for(let ae=U,he=0;ae<Z;ae+=2,he+=D){const Q=ae,B=Q+te,O=B+te,W=O+te,N=M[Q],J=M[Q+1],ie=M[B],me=M[B+1],Ae=M[O],Ve=M[O+1],$e=M[W],X=M[W+1],z=q[he],_e=v*q[he+1],Ee=ie*z-me*_e,Me=ie*_e+me*z,Ce=q[2*he],ye=v*q[2*he+1],de=Ae*Ce-Ve*ye,we=Ae*ye+Ve*Ce,ce=q[3*he],ke=v*q[3*he+1],Le=$e*ce-X*ke,Te=$e*ke+X*ce,We=N+de,qe=J+we,st=N-de,Ze=J-we,ze=Ee+Le,He=Me+Te,gt=v*(Ee-Le),dt=v*(Me-Te);M[Q]=We+ze,M[Q+1]=qe+He,M[B]=st+dt,M[B+1]=Ze-gt,M[O]=We-ze,M[O+1]=qe-He,M[W]=st-dt,M[W+1]=Ze+gt}}}}_singleTransform2(M,T,v,P,F){const D=M[P],K=M[P+1],U=M[P+F],j=M[P+F+1];T[v]=D+U,T[v+1]=K+j,T[v+2]=D-U,T[v+3]=K-j}_singleTransform4(M,T,v,P,F,D){const K=F*2,U=F*3,j=M[P],ne=M[P+1],q=M[P+F],te=M[P+F+1],Z=M[P+K],ae=M[P+K+1],he=M[P+U],Q=M[P+U+1],B=j+Z,O=ne+ae,W=j-Z,N=ne-ae,J=q+he,ie=te+Q,me=D*(q-he),Ae=D*(te-Q);T[v]=B+J,T[v+1]=O+ie,T[v+2]=W+Ae,T[v+3]=N-me,T[v+4]=B-J,T[v+5]=O-ie,T[v+6]=W-Ae,T[v+7]=N+me}_realTransform4(M,T,v){const P=this._csize;let D=1<<this._width,K=P/D<<1,U,j;const ne=this._bitrev;if(K===4)for(U=0,j=0;U<P;U+=K,++j){const Z=ne[j];this._singleRealTransform2(T,M,U,Z>>>1,D>>>1)}else for(U=0,j=0;U<P;U+=K,++j){const Z=ne[j];this._singleRealTransform4(T,M,U,Z>>>1,D>>>1,v)}const q=this.table;for(D>>=2;D>=2;D>>=2){K=P/D<<1;const Z=K>>>1,ae=Z>>>1,he=ae>>>1;for(U=0;U<P;U+=K)for(let Q=0,B=0;Q<=he;Q+=2,B+=D){const O=U+Q,W=O+ae,N=W+ae,J=N+ae,ie=M[O],me=M[O+1],Ae=M[W],Ve=M[W+1],$e=M[N],X=M[N+1],z=M[J],_e=M[J+1],Ee=ie,Me=me,Ce=q[B],ye=v*q[B+1],de=Ae*Ce-Ve*ye,we=Ae*ye+Ve*Ce,ce=q[2*B],ke=v*q[2*B+1],Le=$e*ce-X*ke,Te=$e*ke+X*ce,We=q[3*B],qe=v*q[3*B+1],st=z*We-_e*qe,Ze=z*qe+_e*We,ze=Ee+Le,He=Me+Te,gt=Ee-Le,dt=Me-Te,kt=de+st,ht=we+Ze,yr=v*(de-st),$r=v*(we-Ze);if(M[O]=ze+kt,M[O+1]=He+ht,M[W]=gt+$r,M[W+1]=dt-yr,Q===0){M[N]=ze-kt,M[N+1]=He-ht;continue}if(Q===he)continue;const Vr=U+ae-Q,Ur=U+Z-Q;M[Vr]=gt-v*$r,M[Vr+1]=-dt-v*yr,M[Ur]=ze-v*kt,M[Ur+1]=-He+v*ht}}const te=P>>>1;for(let Z=2;Z<te;Z+=2)M[P-Z]=M[Z],M[P-Z+1]=-M[Z+1]}_singleRealTransform2(M,T,v,P,F){const D=M[P],K=M[P+F];T[v]=D+K,T[v+1]=0,T[v+2]=D-K,T[v+3]=0}_singleRealTransform4(M,T,v,P,F,D){const K=F*2,U=F*3,j=M[P],ne=M[P+F],q=M[P+K],te=M[P+U],Z=j+q,ae=j-q,he=ne+te,Q=D*(ne-te);T[v]=Z+he,T[v+1]=0,T[v+2]=ae,T[v+3]=-Q,T[v+4]=Z-he,T[v+5]=0,T[v+6]=ae,T[v+7]=Q}}class f{constructor(M){const T=2*(M-1),v=2*(2*M-1),P=2**Math.ceil(Math.log2(v));this.bufferSize=P,this._a=T;const F=new Float64Array(v),D=new Float64Array(P);this._chirpBuffer=new Float64Array(P),this._buffer1=new Float64Array(P),this._buffer2=new Float64Array(P),this._outBuffer1=new Float64Array(P),this._outBuffer2=new Float64Array(P);const K=-2*Math.PI/M,U=Math.cos(K),j=Math.sin(K);for(let ne=0;ne<v>>1;++ne){const q=(ne+1-M)**2/2,te=Math.sqrt(U**2+j**2)**q,Z=q*Math.atan2(j,U),ae=2*ne;F[ae]=te*Math.cos(Z),F[ae+1]=te*Math.sin(Z),D[ae]=F[ae],D[ae+1]=-F[ae+1]}this._slicedChirpBuffer=F.subarray(T,v),this._f=new _(P>>1),this._f.transform(this._chirpBuffer,D)}_transform(M,T,v){const P=this._buffer1,F=this._buffer2,D=this._outBuffer1,K=this._outBuffer2,U=this._chirpBuffer,j=this._slicedChirpBuffer,ne=this._a;if(v)for(let q=0;q<j.length;q+=2){const te=q+1,Z=q>>1,ae=T[Z];P[q]=ae*j[q],P[te]=ae*j[te]}else for(let q=0;q<j.length;q+=2){const te=q+1;P[q]=T[q]*j[q]-T[te]*j[te],P[te]=T[q]*j[te]+T[te]*j[q]}this._f.transform(D,P);for(let q=0;q<U.length;q+=2){const te=q+1;F[q]=D[q]*U[q]-D[te]*U[te],F[te]=D[q]*U[te]+D[te]*U[q]}this._f.inverseTransform(K,F);for(let q=0;q<K.length;q+=2){const te=K[q+ne],Z=K[q+ne+1],ae=j[q],he=j[q+1];M[q]=te*ae-Z*he,M[q+1]=te*he+Z*ae}}transform(M,T){this._transform(M,T,!1)}realTransform(M,T){this._transform(M,T,!0)}}class b{constructor(M){this.fft_length=M,this.isPowerOfTwo=u(M),this.isPowerOfTwo?(this.fft=new _(M),this.outputBufferSize=2*M):(this.fft=new f(M),this.outputBufferSize=this.fft.bufferSize)}realTransform(M,T){this.fft.realTransform(M,T)}transform(M,T){this.fft.transform(M,T)}}function A(x,M){if(M%2===0||M<=0)throw new Error("Window size must be a positive odd number");const T=new x.constructor(x.length),v=new x.constructor(M),P=Math.floor(M/2);for(let F=0;F<x.length;++F){let D=0;for(let K=-P;K<=P;++K){let U=F+K;U<0?U=Math.abs(U):U>=x.length&&(U=2*(x.length-1)-U),v[D++]=x[U]}v.sort(),T[F]=v[P]}return T}function g(x,M){const T=Math.pow(10,M);return Math.round(x*T)/T}function y(x){const M=Math.round(x);return Math.abs(x)%1===.5?M%2===0?M:M-1:M}function C(x){const M=x.length,T=x[0].length,v=[M+1,T+1],P=Array.from({length:v[0]},()=>Array(v[1]).fill(1/0));P[0][0]=0;const F=Array.from({length:v[0]},()=>Array(v[1]).fill(-1));for(let ne=1;ne<v[1];++ne)for(let q=1;q<v[0];++q){const te=P[q-1][ne-1],Z=P[q-1][ne],ae=P[q][ne-1];let he,Q;te<Z&&te<ae?(he=te,Q=0):Z<te&&Z<ae?(he=Z,Q=1):(he=ae,Q=2),P[q][ne]=x[q-1][ne-1]+he,F[q][ne]=Q}for(let ne=0;ne<v[1];++ne)F[0][ne]=2;for(let ne=0;ne<v[0];++ne)F[ne][0]=1;let D=M,K=T,U=[],j=[];for(;D>0||K>0;)switch(U.push(D-1),j.push(K-1),F[D][K]){case 0:--D,--K;break;case 1:--D;break;case 2:--K;break;default:throw new Error(`Internal error in dynamic time warping. Unexpected trace[${D}, ${K}]. Please file a bug report.`)}return U.reverse(),j.reverse(),[U,j]}},"./src/utils/tensor.js":(e,r,t)=>{t.r(r),t.d(r,{DataTypeMap:()=>a,Tensor:()=>i,cat:()=>T,full:()=>j,full_like:()=>ne,interpolate:()=>p,interpolate_4d:()=>d,layer_norm:()=>y,matmul:()=>u,mean:()=>D,mean_pooling:()=>g,ones:()=>q,ones_like:()=>te,permute:()=>c,quantize_embeddings:()=>Q,rand:()=>he,rfft:()=>_,slice:()=>A,stack:()=>v,std_mean:()=>F,topk:()=>f,zeros:()=>Z,zeros_like:()=>ae});var s=t("./src/utils/maths.js"),o=t("./src/backends/onnx.js"),n=t("./src/ops/registry.js");const a=Object.freeze({float32:Float32Array,float16:typeof Float16Array<"u"?Float16Array:Uint16Array,float64:Float64Array,string:Array,int8:Int8Array,uint8:Uint8Array,int16:Int16Array,uint16:Uint16Array,int32:Int32Array,uint32:Uint32Array,int64:BigInt64Array,uint64:BigUint64Array,bool:Uint8Array,uint4:Uint8Array,int4:Int8Array});class i{constructor(...O){re(this,"ort_tensor");return(0,o.isONNXTensor)(O[0])?this.ort_tensor=O[0]:this.ort_tensor=new o.Tensor(O[0],O[1],O[2]),new Proxy(this,{get:(W,N)=>{if(typeof N=="string"){let J=Number(N);if(Number.isInteger(J))return W._getitem(J)}return W[N]},set:(W,N,J)=>W[N]=J})}get dims(){return this.ort_tensor.dims}set dims(O){this.ort_tensor.dims=O}get type(){return this.ort_tensor.type}get data(){return this.ort_tensor.data}get size(){return this.ort_tensor.size}get location(){return this.ort_tensor.location}dispose(){this.ort_tensor.dispose()}*[Symbol.iterator](){const[O,...W]=this.dims;if(W.length>0){const N=W.reduce((J,ie)=>J*ie);for(let J=0;J<O;++J)yield this._subarray(J,N,W)}else yield*this.data}_getitem(O){const[W,...N]=this.dims;if(O=M(O,W),N.length>0){const J=N.reduce((ie,me)=>ie*me);return this._subarray(O,J,N)}else return new i(this.type,[this.data[O]],N)}indexOf(O){const W=this.data;for(let N=0;N<W.length;++N)if(W[N]==O)return N;return-1}_subarray(O,W,N){const J=O*W,ie=(O+1)*W,me="subarray"in this.data?this.data.subarray(J,ie):this.data.slice(J,ie);return new i(this.type,me,N)}item(){const O=this.data;if(O.length!==1)throw new Error(`a Tensor with ${O.length} elements cannot be converted to Scalar`);return O[0]}tolist(){return l(this.data,this.dims)}sigmoid(){return this.clone().sigmoid_()}sigmoid_(){const O=this.data;for(let W=0;W<O.length;++W)O[W]=1/(1+Math.exp(-O[W]));return this}map(O){return this.clone().map_(O)}map_(O){const W=this.data;for(let N=0;N<W.length;++N)W[N]=O(W[N],N,W);return this}mul(O){return this.clone().mul_(O)}mul_(O){const W=this.data;for(let N=0;N<W.length;++N)W[N]*=O;return this}div(O){return this.clone().div_(O)}div_(O){const W=this.data;for(let N=0;N<W.length;++N)W[N]/=O;return this}add(O){return this.clone().add_(O)}add_(O){const W=this.data;for(let N=0;N<W.length;++N)W[N]+=O;return this}sub(O){return this.clone().sub_(O)}sub_(O){const W=this.data;for(let N=0;N<W.length;++N)W[N]-=O;return this}clone(){return new i(this.type,this.data.slice(),this.dims.slice())}slice(...O){const W=[],N=[];for(let X=0;X<this.dims.length;++X){let z=O[X];if(z==null)N.push([0,this.dims[X]]),W.push(this.dims[X]);else if(typeof z=="number")z=M(z,this.dims[X],X),N.push([z,z+1]);else if(Array.isArray(z)&&z.length===2){let[_e,Ee]=z;if(_e=_e===null?0:M(_e,this.dims[X],X,!1),Ee=Ee===null?this.dims[X]:M(Ee,this.dims[X],X,!1),_e>Ee)throw new Error(`Invalid slice: ${z}`);const Me=[Math.max(_e,0),Math.min(Ee,this.dims[X])];N.push(Me),W.push(Me[1]-Me[0])}else throw new Error(`Invalid slice: ${z}`)}const J=N.map(([X,z])=>z-X),ie=J.reduce((X,z)=>X*z),me=this.data,Ae=new me.constructor(ie),Ve=this.stride();let $e=!0;for(let X=1;X<J.length;++X)if(N[X][0]!==0||N[X][1]!==this.dims[X]){$e=!1;break}if($e){const X=N[0][0]*Ve[0],z=N[0][1]*Ve[0];if(ArrayBuffer.isView(me))Ae.set(me.subarray(X,z));else if(Array.isArray(me)){const _e=me.slice(X,z);for(let Ee=0;Ee<_e.length;++Ee)Ae[Ee]=_e[Ee]}else throw new Error("Unsupported data type for slicing")}else for(let X=0;X<ie;++X){let z=0;for(let _e=J.length-1,Ee=X;_e>=0;--_e){const Me=J[_e];z+=(Ee%Me+N[_e][0])*Ve[_e],Ee=Math.floor(Ee/Me)}Ae[X]=me[z]}return new i(this.type,Ae,W)}permute(...O){return c(this,O)}transpose(...O){return this.permute(...O)}sum(O=null,W=!1){return this.norm(1,O,W)}norm(O="fro",W=null,N=!1){if(O==="fro")O=2;else if(typeof O=="string")throw Error(`Unsupported norm: ${O}`);const J=this.data,ie=($e,X)=>$e+X**O;if(W===null){const $e=J.reduce(ie,0)**(1/O);return new i(this.type,[$e],[])}const[me,Ae,Ve]=P(ie,this,W,N);if(O!==1)for(let $e=0;$e<Ae.length;++$e)Ae[$e]=Ae[$e]**(1/O);return new i(me,Ae,Ve)}normalize_(O=2,W=1){W=M(W,this.dims.length);const N=this.norm(O,W,!0),J=this.data,ie=N.data;for(let me=0;me<J.length;++me){let Ae=0;for(let Ve=this.dims.length-1,$e=me,X=1;Ve>=0;--Ve){const z=this.dims[Ve];if(Ve!==W){const _e=$e%z;Ae+=_e*X,X*=this.dims[Ve]}$e=Math.floor($e/z)}J[me]/=ie[Ae]}return this}normalize(O=2,W=1){return this.clone().normalize_(O,W)}stride(){return K(this.dims)}squeeze(O=null){return new i(this.type,this.data,C(this.dims,O))}squeeze_(O=null){return this.dims=C(this.dims,O),this}unsqueeze(O=null){return new i(this.type,this.data,x(this.dims,O))}unsqueeze_(O=null){return this.dims=x(this.dims,O),this}flatten_(O=0,W=-1){W=(W+this.dims.length)%this.dims.length;let N=this.dims.slice(0,O),J=this.dims.slice(O,W+1),ie=this.dims.slice(W+1);return this.dims=[...N,J.reduce((me,Ae)=>me*Ae,1),...ie],this}flatten(O=0,W=-1){return this.clone().flatten_(O,W)}view(...O){let W=-1;for(let J=0;J<O.length;++J)if(O[J]===-1){if(W!==-1)throw new Error("Only one dimension can be inferred");W=J}const N=this.data;if(W!==-1){const J=O.reduce((ie,me,Ae)=>Ae!==W?ie*me:ie,1);O[W]=N.length/J}return new i(this.type,N,O)}neg_(){const O=this.data;for(let W=0;W<O.length;++W)O[W]=-O[W];return this}neg(){return this.clone().neg_()}gt(O){const W=new Uint8Array(this.data.length),N=this.data;for(let J=0;J<N.length;++J)W[J]=N[J]>O?1:0;return new i("bool",W,this.dims)}lt(O){const W=new Uint8Array(this.data.length),N=this.data;for(let J=0;J<N.length;++J)W[J]=N[J]<O?1:0;return new i("bool",W,this.dims)}clamp_(O,W){const N=this.data;for(let J=0;J<N.length;++J)N[J]=Math.min(Math.max(N[J],O),W);return this}clamp(O,W){return this.clone().clamp_(O,W)}round_(){const O=this.data;for(let W=0;W<O.length;++W)O[W]=Math.round(O[W]);return this}round(){return this.clone().round_()}mean(O=null,W=!1){return D(this,O,W)}min(O=null,W=!1){if(O===null){const me=(0,s.min)(this.data)[0];return new i(this.type,[me],[])}const[N,J,ie]=P((me,Ae)=>Math.min(me,Ae),this,O,W,1/0);return new i(N,J,ie)}max(O=null,W=!1){if(O===null){const me=(0,s.max)(this.data)[0];return new i(this.type,[me],[])}const[N,J,ie]=P((me,Ae)=>Math.max(me,Ae),this,O,W,-1/0);return new i(N,J,ie)}argmin(O=null,W=!1){if(O!==null)throw new Error("`dim !== null` not yet implemented.");const N=(0,s.min)(this.data)[1];return new i("int64",[BigInt(N)],[])}argmax(O=null,W=!1){if(O!==null)throw new Error("`dim !== null` not yet implemented.");const N=(0,s.max)(this.data)[1];return new i("int64",[BigInt(N)],[])}to(O){if(this.type===O)return this;if(!a.hasOwnProperty(O))throw new Error(`Unsupported type: ${O}`);let W;const N=["int64","uint64"].includes(this.type),J=["int64","uint64"].includes(O);return N&&!J?W=Number:!N&&J&&(W=BigInt),new i(O,a[O].from(this.data,W),this.dims)}}function l(B,O){const W=B.length,N=O.reduce((ie,me)=>ie*me);if(W!==N)throw Error(`cannot reshape array of size ${W} into shape (${O})`);let J=B;for(let ie=O.length-1;ie>=0;ie--)J=J.reduce((me,Ae)=>{let Ve=me[me.length-1];return Ve.length<O[ie]?Ve.push(Ae):me.push([Ae]),me},[[]]);return J[0]}function c(B,O){const[W,N]=(0,s.permute_data)(B.data,B.dims,O);return new i(B.type,W,N)}function p(B,[O,W],N="bilinear",J=!1){const ie=B.dims.at(-3)??1,me=B.dims.at(-2),Ae=B.dims.at(-1);let Ve=(0,s.interpolate_data)(B.data,[ie,me,Ae],[O,W],N,J);return new i(B.type,Ve,[ie,O,W])}async function d(B,{size:O=null,mode:W="bilinear"}={}){if(B.dims.length!==4)throw new Error("`interpolate_4d` currently only supports 4D input.");if(!O)throw new Error("`interpolate_4d` requires a `size` argument.");let N;if(O.length===2)N=[...B.dims.slice(0,2),...O];else if(O.length===3)N=[B.dims[0],...O];else if(O.length===4)N=O;else throw new Error("`size` must be of length 2, 3, or 4.");let J;if(W==="nearest")J=await n.TensorOpRegistry.nearest_interpolate_4d;else if(W==="bilinear")J=await n.TensorOpRegistry.bilinear_interpolate_4d;else if(W==="bicubic")J=await n.TensorOpRegistry.bicubic_interpolate_4d;else throw new Error(`Unsupported mode: ${W}`);const ie=new i("int64",new BigInt64Array(N.map(BigInt)),[N.length]);return await J({x:B,s:ie})}async function u(B,O){return await(await n.TensorOpRegistry.matmul)({a:B,b:O})}async function _(B,O){return await(await n.TensorOpRegistry.rfft)({x:B,a:O})}async function f(B,O){const W=await n.TensorOpRegistry.top_k;return O==null?O=B.dims.at(-1):O=Math.min(O,B.dims.at(-1)),await W({x:B,k:new i("int64",[BigInt(O)],[1])})}const b=B=>new i("int64",B,[B.length]);async function A(B,O,W,N,J){return await(await n.TensorOpRegistry.slice)({x:B,s:b(O),e:b(W),a:b(N),t:b(J??new Array(N.length).fill(1))})}function g(B,O){const W=B.data,N=O.data,J=[B.dims[0],B.dims[2]],ie=new W.constructor(J[0]*J[1]),[me,Ae,Ve]=B.dims;let $e=0;for(let X=0;X<me;++X){const z=X*Ve*Ae;for(let _e=0;_e<Ve;++_e){let Ee=0,Me=0;const Ce=X*Ae,ye=z+_e;for(let we=0;we<Ae;++we){const ce=Number(N[Ce+we]);Me+=ce,Ee+=W[ye+we*Ve]*ce}const de=Ee/Me;ie[$e++]=de}}return new i(B.type,ie,J)}function y(B,O,{eps:W=1e-5}={}){if(B.dims.length!==2)throw new Error("`layer_norm` currently only supports 2D input.");const[N,J]=B.dims;if(O.length!==1&&O[0]!==J)throw new Error("`normalized_shape` must be a 1D array with shape `[input.dims[1]]`.");const[ie,me]=F(B,1,0,!0),Ae=ie.data,Ve=me.data,$e=B.data,X=new $e.constructor($e.length);for(let z=0;z<N;++z){const _e=z*J;for(let Ee=0;Ee<J;++Ee){const Me=_e+Ee;X[Me]=($e[Me]-Ve[z])/(Ae[z]+W)}}return new i(B.type,X,B.dims)}function C(B,O){return B=B.slice(),O===null?B=B.filter(W=>W!==1):typeof O=="number"?B[O]===1&&B.splice(O,1):Array.isArray(O)&&(B=B.filter((W,N)=>W!==1||!O.includes(N))),B}function x(B,O){return O=M(O,B.length+1),B=B.slice(),B.splice(O,0,1),B}function M(B,O,W=null,N=!0){if(B<-O||B>=O){if(N)throw new Error(`IndexError: index ${B} is out of bounds for dimension${W===null?"":" "+W} with size ${O}`);return B<-O?0:O}return B<0&&(B=(B%O+O)%O),B}function T(B,O=0){O=M(O,B[0].dims.length);const W=B[0].dims.slice();W[O]=B.reduce((me,Ae)=>me+Ae.dims[O],0);const N=W.reduce((me,Ae)=>me*Ae,1),J=new B[0].data.constructor(N),ie=B[0].type;if(O===0){let me=0;for(const Ae of B){const Ve=Ae.data;J.set(Ve,me),me+=Ve.length}}else{let me=0;for(let Ae=0;Ae<B.length;++Ae){const{data:Ve,dims:$e}=B[Ae];for(let X=0;X<Ve.length;++X){let z=0;for(let _e=$e.length-1,Ee=X,Me=1;_e>=0;--_e){const Ce=$e[_e];let ye=Ee%Ce;_e===O&&(ye+=me),z+=ye*Me,Me*=W[_e],Ee=Math.floor(Ee/Ce)}J[z]=Ve[X]}me+=$e[O]}}return new i(ie,J,W)}function v(B,O=0){return T(B.map(W=>W.unsqueeze(O)),O)}function P(B,O,W=null,N=!1,J=null){const ie=O.data,me=O.dims;W=M(W,me.length);const Ae=me.slice();Ae[W]=1;const Ve=new ie.constructor(ie.length/me[W]);J!==null&&Ve.fill(J);for(let $e=0;$e<ie.length;++$e){let X=0;for(let z=me.length-1,_e=$e,Ee=1;z>=0;--z){const Me=me[z];if(z!==W){const Ce=_e%Me;X+=Ce*Ee,Ee*=Ae[z]}_e=Math.floor(_e/Me)}Ve[X]=B(Ve[X],ie[$e],$e,X)}return N||Ae.splice(W,1),[O.type,Ve,Ae]}function F(B,O=null,W=1,N=!1){const J=B.data,ie=B.dims;if(O===null){const Ee=J.reduce((de,we)=>de+we,0)/J.length,Me=Math.sqrt(J.reduce((de,we)=>de+(we-Ee)**2,0)/(J.length-W)),Ce=new i(B.type,[Ee],[]);return[new i(B.type,[Me],[]),Ce]}O=M(O,ie.length);const me=D(B,O,N),Ae=me.data,[Ve,$e,X]=P((_e,Ee,Me,Ce)=>_e+(Ee-Ae[Ce])**2,B,O,N);for(let _e=0;_e<$e.length;++_e)$e[_e]=Math.sqrt($e[_e]/(ie[O]-W));return[new i(Ve,$e,X),me]}function D(B,O=null,W=!1){const N=B.dims,J=B.data;if(O===null){const Ve=J.reduce(($e,X)=>$e+X,0);return new i(B.type,[Ve/J.length],[])}O=M(O,N.length);const[ie,me,Ae]=P((Ve,$e)=>Ve+$e,B,O,W);if(N[O]!==1)for(let Ve=0;Ve<me.length;++Ve)me[Ve]/=N[O];return new i(ie,me,Ae)}function K(B){const O=new Array(B.length);for(let W=B.length-1,N=1;W>=0;--W)O[W]=N,N*=B[W];return O}function U(B,O,W,N){const J=B.reduce((ie,me)=>ie*me,1);return new i(W,new N(J).fill(O),B)}function j(B,O){let W,N;if(typeof O=="number")W="float32",N=Float32Array;else if(typeof O=="bigint")W="int64",N=BigInt64Array;else if(typeof O=="boolean")W="bool",N=Uint8Array;else throw new Error(`Unsupported data type: ${typeof O}`);return U(B,O,W,N)}function ne(B,O){return j(B.dims,O)}function q(B){return U(B,1n,"int64",BigInt64Array)}function te(B){return q(B.dims)}function Z(B){return U(B,0n,"int64",BigInt64Array)}function ae(B){return Z(B.dims)}function he(B){const O=B.reduce((W,N)=>W*N,1);return new i("float32",Float32Array.from({length:O},()=>Math.random()),B)}function Q(B,O){if(B.dims.length!==2)throw new Error("The tensor must have 2 dimensions");if(B.dims.at(-1)%8!==0)throw new Error("The last dimension of the tensor must be a multiple of 8");if(!["binary","ubinary"].includes(O))throw new Error("The precision must be either 'binary' or 'ubinary'");const W=O==="binary",N=W?"int8":"uint8",J=W?Int8Array:Uint8Array,ie=B.data,me=new J(ie.length/8);for(let Ae=0;Ae<ie.length;++Ae){const Ve=ie[Ae]>0?1:0,$e=Math.floor(Ae/8),X=Ae%8;me[$e]|=Ve<<7-X,W&&X===0&&(me[$e]-=128)}return new i(N,me,[B.dims[0],B.dims[1]/8])}},"./src/utils/video.js":(e,r,t)=>{t.r(r),t.d(r,{RawVideo:()=>a,RawVideoFrame:()=>n,load_video:()=>i});var s=t("./src/utils/image.js"),o=t("./src/env.js");class n{constructor(c,p){this.image=c,this.timestamp=p}}class a{constructor(c,p){c.length>0&&c[0]instanceof s.RawImage&&(c=c.map((d,u)=>new n(d,(u+1)/(c.length+1)*p))),this.frames=c,this.duration=p}get width(){return this.frames[0].image.width}get height(){return this.frames[0].image.height}get fps(){return this.frames.length/this.duration}}async function i(l,{num_frames:c=null,fps:p=null}={}){if(!o.apis.IS_BROWSER_ENV)throw new Error("`load_video` is currently only supported in browser environments.");if(c==null&&p==null)throw new Error("Either num_frames or fps must be provided.");const d=[],u=document.createElement("video");if(u.crossOrigin="anonymous",u.muted=!0,typeof l=="string")u.src=l;else if(l instanceof Blob)u.src=URL.createObjectURL(l);else if(l instanceof HTMLVideoElement)u.src=l.src;else throw new Error("Invalid URL or video element provided.");if(await new Promise(C=>u.onloadedmetadata=C),u.seekable.start(0)===u.seekable.end(0)){const x=await(await fetch(u.src)).blob();u.src=URL.createObjectURL(x),await new Promise(M=>u.onloadedmetadata=M)}const _=u.duration;let f,b;c!=null?(f=c,b=c===1?0:_/(c-1)):(b=1/p,f=Math.floor(_/b));let A=[];for(let C=0;C<f;++C)A.push(c===1?_/2:C*b);const g=document.createElement("canvas");g.width=u.videoWidth,g.height=u.videoHeight;const y=g.getContext("2d",{willReadFrequently:!0});for(const C of A){u.currentTime=C,await new Promise(v=>{u.onseeked=v}),y.drawImage(u,0,0,g.width,g.height);const x=y.getImageData(0,0,g.width,g.height),M=new s.RawImage(x.data,g.width,g.height,4),T=new n(M,C);d.push(T)}return u.remove(),new a(d,_)}}},gw={};function Vt(e){var r=gw[e];if(r!==void 0)return r.exports;var t=gw[e]={exports:{}};return WT[e](t,t.exports,Vt),t.exports}(()=>{var e=Object.getPrototypeOf?t=>Object.getPrototypeOf(t):t=>t.__proto__,r;Vt.t=function(t,s){if(s&1&&(t=this(t)),s&8||typeof t=="object"&&t&&(s&4&&t.__esModule||s&16&&typeof t.then=="function"))return t;var o=Object.create(null);Vt.r(o);var n={};r=r||[null,e({}),e([]),e(e)];for(var a=s&2&&t;typeof a=="object"&&!~r.indexOf(a);a=e(a))Object.getOwnPropertyNames(a).forEach(i=>n[i]=()=>t[i]);return n.default=()=>t,Vt.d(o,n),o}})(),Vt.d=(e,r)=>{for(var t in r)Vt.o(r,t)&&!Vt.o(e,t)&&Object.defineProperty(e,t,{enumerable:!0,get:r[t]})},Vt.o=(e,r)=>Object.prototype.hasOwnProperty.call(e,r),Vt.r=e=>{typeof Symbol<"u"&&Symbol.toStringTag&&Object.defineProperty(e,Symbol.toStringTag,{value:"Module"}),Object.defineProperty(e,"__esModule",{value:!0})};var m={};(()=>{/*!*****************************!*\
  !*** ./src/transformers.js ***!
  \*****************************/Vt.r(m),Vt.d(m,{ASTFeatureExtractor:()=>d.ASTFeatureExtractor,ASTForAudioClassification:()=>t.ASTForAudioClassification,ASTModel:()=>t.ASTModel,ASTPreTrainedModel:()=>t.ASTPreTrainedModel,AlbertForMaskedLM:()=>t.AlbertForMaskedLM,AlbertForQuestionAnswering:()=>t.AlbertForQuestionAnswering,AlbertForSequenceClassification:()=>t.AlbertForSequenceClassification,AlbertModel:()=>t.AlbertModel,AlbertPreTrainedModel:()=>t.AlbertPreTrainedModel,AlbertTokenizer:()=>s.AlbertTokenizer,ArceeForCausalLM:()=>t.ArceeForCausalLM,ArceeModel:()=>t.ArceeModel,ArceePreTrainedModel:()=>t.ArceePreTrainedModel,AudioClassificationPipeline:()=>r.AudioClassificationPipeline,AutoConfig:()=>o.AutoConfig,AutoFeatureExtractor:()=>u.AutoFeatureExtractor,AutoImageProcessor:()=>b.AutoImageProcessor,AutoModel:()=>t.AutoModel,AutoModelForAudioClassification:()=>t.AutoModelForAudioClassification,AutoModelForAudioFrameClassification:()=>t.AutoModelForAudioFrameClassification,AutoModelForAudioTextToText:()=>t.AutoModelForAudioTextToText,AutoModelForCTC:()=>t.AutoModelForCTC,AutoModelForCausalLM:()=>t.AutoModelForCausalLM,AutoModelForDepthEstimation:()=>t.AutoModelForDepthEstimation,AutoModelForDocumentQuestionAnswering:()=>t.AutoModelForDocumentQuestionAnswering,AutoModelForImageClassification:()=>t.AutoModelForImageClassification,AutoModelForImageFeatureExtraction:()=>t.AutoModelForImageFeatureExtraction,AutoModelForImageMatting:()=>t.AutoModelForImageMatting,AutoModelForImageSegmentation:()=>t.AutoModelForImageSegmentation,AutoModelForImageTextToText:()=>t.AutoModelForImageTextToText,AutoModelForImageToImage:()=>t.AutoModelForImageToImage,AutoModelForMaskGeneration:()=>t.AutoModelForMaskGeneration,AutoModelForMaskedLM:()=>t.AutoModelForMaskedLM,AutoModelForNormalEstimation:()=>t.AutoModelForNormalEstimation,AutoModelForObjectDetection:()=>t.AutoModelForObjectDetection,AutoModelForPoseEstimation:()=>t.AutoModelForPoseEstimation,AutoModelForQuestionAnswering:()=>t.AutoModelForQuestionAnswering,AutoModelForSemanticSegmentation:()=>t.AutoModelForSemanticSegmentation,AutoModelForSeq2SeqLM:()=>t.AutoModelForSeq2SeqLM,AutoModelForSequenceClassification:()=>t.AutoModelForSequenceClassification,AutoModelForSpeechSeq2Seq:()=>t.AutoModelForSpeechSeq2Seq,AutoModelForTextToSpectrogram:()=>t.AutoModelForTextToSpectrogram,AutoModelForTextToWaveform:()=>t.AutoModelForTextToWaveform,AutoModelForTokenClassification:()=>t.AutoModelForTokenClassification,AutoModelForUniversalSegmentation:()=>t.AutoModelForUniversalSegmentation,AutoModelForVision2Seq:()=>t.AutoModelForVision2Seq,AutoModelForXVector:()=>t.AutoModelForXVector,AutoModelForZeroShotObjectDetection:()=>t.AutoModelForZeroShotObjectDetection,AutoProcessor:()=>y.AutoProcessor,AutoTokenizer:()=>s.AutoTokenizer,AutomaticSpeechRecognitionPipeline:()=>r.AutomaticSpeechRecognitionPipeline,BackgroundRemovalPipeline:()=>r.BackgroundRemovalPipeline,BartForConditionalGeneration:()=>t.BartForConditionalGeneration,BartForSequenceClassification:()=>t.BartForSequenceClassification,BartModel:()=>t.BartModel,BartPretrainedModel:()=>t.BartPretrainedModel,BartTokenizer:()=>s.BartTokenizer,BaseModelOutput:()=>t.BaseModelOutput,BaseStreamer:()=>C.BaseStreamer,BeitFeatureExtractor:()=>f.BeitFeatureExtractor,BeitForImageClassification:()=>t.BeitForImageClassification,BeitModel:()=>t.BeitModel,BeitPreTrainedModel:()=>t.BeitPreTrainedModel,BertForMaskedLM:()=>t.BertForMaskedLM,BertForQuestionAnswering:()=>t.BertForQuestionAnswering,BertForSequenceClassification:()=>t.BertForSequenceClassification,BertForTokenClassification:()=>t.BertForTokenClassification,BertModel:()=>t.BertModel,BertPreTrainedModel:()=>t.BertPreTrainedModel,BertTokenizer:()=>s.BertTokenizer,BitImageProcessor:()=>f.BitImageProcessor,BlenderbotForConditionalGeneration:()=>t.BlenderbotForConditionalGeneration,BlenderbotModel:()=>t.BlenderbotModel,BlenderbotPreTrainedModel:()=>t.BlenderbotPreTrainedModel,BlenderbotSmallForConditionalGeneration:()=>t.BlenderbotSmallForConditionalGeneration,BlenderbotSmallModel:()=>t.BlenderbotSmallModel,BlenderbotSmallPreTrainedModel:()=>t.BlenderbotSmallPreTrainedModel,BlenderbotSmallTokenizer:()=>s.BlenderbotSmallTokenizer,BlenderbotTokenizer:()=>s.BlenderbotTokenizer,BloomForCausalLM:()=>t.BloomForCausalLM,BloomModel:()=>t.BloomModel,BloomPreTrainedModel:()=>t.BloomPreTrainedModel,BloomTokenizer:()=>s.BloomTokenizer,CLIPFeatureExtractor:()=>f.CLIPFeatureExtractor,CLIPImageProcessor:()=>f.CLIPImageProcessor,CLIPModel:()=>t.CLIPModel,CLIPPreTrainedModel:()=>t.CLIPPreTrainedModel,CLIPSegForImageSegmentation:()=>t.CLIPSegForImageSegmentation,CLIPSegModel:()=>t.CLIPSegModel,CLIPSegPreTrainedModel:()=>t.CLIPSegPreTrainedModel,CLIPTextModel:()=>t.CLIPTextModel,CLIPTextModelWithProjection:()=>t.CLIPTextModelWithProjection,CLIPTokenizer:()=>s.CLIPTokenizer,CLIPVisionModel:()=>t.CLIPVisionModel,CLIPVisionModelWithProjection:()=>t.CLIPVisionModelWithProjection,CamembertForMaskedLM:()=>t.CamembertForMaskedLM,CamembertForQuestionAnswering:()=>t.CamembertForQuestionAnswering,CamembertForSequenceClassification:()=>t.CamembertForSequenceClassification,CamembertForTokenClassification:()=>t.CamembertForTokenClassification,CamembertModel:()=>t.CamembertModel,CamembertPreTrainedModel:()=>t.CamembertPreTrainedModel,CamembertTokenizer:()=>s.CamembertTokenizer,CausalLMOutput:()=>t.CausalLMOutput,CausalLMOutputWithPast:()=>t.CausalLMOutputWithPast,ChineseCLIPFeatureExtractor:()=>f.ChineseCLIPFeatureExtractor,ChineseCLIPModel:()=>t.ChineseCLIPModel,ChineseCLIPPreTrainedModel:()=>t.ChineseCLIPPreTrainedModel,ClapAudioModelWithProjection:()=>t.ClapAudioModelWithProjection,ClapFeatureExtractor:()=>d.ClapFeatureExtractor,ClapModel:()=>t.ClapModel,ClapPreTrainedModel:()=>t.ClapPreTrainedModel,ClapTextModelWithProjection:()=>t.ClapTextModelWithProjection,ClassifierFreeGuidanceLogitsProcessor:()=>M.ClassifierFreeGuidanceLogitsProcessor,CodeGenForCausalLM:()=>t.CodeGenForCausalLM,CodeGenModel:()=>t.CodeGenModel,CodeGenPreTrainedModel:()=>t.CodeGenPreTrainedModel,CodeGenTokenizer:()=>s.CodeGenTokenizer,CodeLlamaTokenizer:()=>s.CodeLlamaTokenizer,CohereForCausalLM:()=>t.CohereForCausalLM,CohereModel:()=>t.CohereModel,CoherePreTrainedModel:()=>t.CoherePreTrainedModel,CohereTokenizer:()=>s.CohereTokenizer,ConvBertForMaskedLM:()=>t.ConvBertForMaskedLM,ConvBertForQuestionAnswering:()=>t.ConvBertForQuestionAnswering,ConvBertForSequenceClassification:()=>t.ConvBertForSequenceClassification,ConvBertForTokenClassification:()=>t.ConvBertForTokenClassification,ConvBertModel:()=>t.ConvBertModel,ConvBertPreTrainedModel:()=>t.ConvBertPreTrainedModel,ConvBertTokenizer:()=>s.ConvBertTokenizer,ConvNextFeatureExtractor:()=>f.ConvNextFeatureExtractor,ConvNextForImageClassification:()=>t.ConvNextForImageClassification,ConvNextImageProcessor:()=>f.ConvNextImageProcessor,ConvNextModel:()=>t.ConvNextModel,ConvNextPreTrainedModel:()=>t.ConvNextPreTrainedModel,ConvNextV2ForImageClassification:()=>t.ConvNextV2ForImageClassification,ConvNextV2Model:()=>t.ConvNextV2Model,ConvNextV2PreTrainedModel:()=>t.ConvNextV2PreTrainedModel,DFineForObjectDetection:()=>t.DFineForObjectDetection,DFineModel:()=>t.DFineModel,DFinePreTrainedModel:()=>t.DFinePreTrainedModel,DINOv3ConvNextModel:()=>t.DINOv3ConvNextModel,DINOv3ConvNextPreTrainedModel:()=>t.DINOv3ConvNextPreTrainedModel,DINOv3ViTImageProcessor:()=>f.DINOv3ViTImageProcessor,DINOv3ViTModel:()=>t.DINOv3ViTModel,DINOv3ViTPreTrainedModel:()=>t.DINOv3ViTPreTrainedModel,DPTFeatureExtractor:()=>f.DPTFeatureExtractor,DPTForDepthEstimation:()=>t.DPTForDepthEstimation,DPTImageProcessor:()=>f.DPTImageProcessor,DPTModel:()=>t.DPTModel,DPTPreTrainedModel:()=>t.DPTPreTrainedModel,DacDecoderModel:()=>t.DacDecoderModel,DacDecoderOutput:()=>t.DacDecoderOutput,DacEncoderModel:()=>t.DacEncoderModel,DacEncoderOutput:()=>t.DacEncoderOutput,DacFeatureExtractor:()=>d.DacFeatureExtractor,DacModel:()=>t.DacModel,DacPreTrainedModel:()=>t.DacPreTrainedModel,DataTypeMap:()=>l.DataTypeMap,DebertaForMaskedLM:()=>t.DebertaForMaskedLM,DebertaForQuestionAnswering:()=>t.DebertaForQuestionAnswering,DebertaForSequenceClassification:()=>t.DebertaForSequenceClassification,DebertaForTokenClassification:()=>t.DebertaForTokenClassification,DebertaModel:()=>t.DebertaModel,DebertaPreTrainedModel:()=>t.DebertaPreTrainedModel,DebertaTokenizer:()=>s.DebertaTokenizer,DebertaV2ForMaskedLM:()=>t.DebertaV2ForMaskedLM,DebertaV2ForQuestionAnswering:()=>t.DebertaV2ForQuestionAnswering,DebertaV2ForSequenceClassification:()=>t.DebertaV2ForSequenceClassification,DebertaV2ForTokenClassification:()=>t.DebertaV2ForTokenClassification,DebertaV2Model:()=>t.DebertaV2Model,DebertaV2PreTrainedModel:()=>t.DebertaV2PreTrainedModel,DebertaV2Tokenizer:()=>s.DebertaV2Tokenizer,DecisionTransformerModel:()=>t.DecisionTransformerModel,DecisionTransformerPreTrainedModel:()=>t.DecisionTransformerPreTrainedModel,DeiTFeatureExtractor:()=>f.DeiTFeatureExtractor,DeiTForImageClassification:()=>t.DeiTForImageClassification,DeiTImageProcessor:()=>f.DeiTImageProcessor,DeiTModel:()=>t.DeiTModel,DeiTPreTrainedModel:()=>t.DeiTPreTrainedModel,DepthAnythingForDepthEstimation:()=>t.DepthAnythingForDepthEstimation,DepthAnythingPreTrainedModel:()=>t.DepthAnythingPreTrainedModel,DepthEstimationPipeline:()=>r.DepthEstimationPipeline,DepthProForDepthEstimation:()=>t.DepthProForDepthEstimation,DepthProPreTrainedModel:()=>t.DepthProPreTrainedModel,DetrFeatureExtractor:()=>f.DetrFeatureExtractor,DetrForObjectDetection:()=>t.DetrForObjectDetection,DetrForSegmentation:()=>t.DetrForSegmentation,DetrImageProcessor:()=>f.DetrImageProcessor,DetrModel:()=>t.DetrModel,DetrObjectDetectionOutput:()=>t.DetrObjectDetectionOutput,DetrPreTrainedModel:()=>t.DetrPreTrainedModel,DetrSegmentationOutput:()=>t.DetrSegmentationOutput,Dinov2ForImageClassification:()=>t.Dinov2ForImageClassification,Dinov2Model:()=>t.Dinov2Model,Dinov2PreTrainedModel:()=>t.Dinov2PreTrainedModel,Dinov2WithRegistersForImageClassification:()=>t.Dinov2WithRegistersForImageClassification,Dinov2WithRegistersModel:()=>t.Dinov2WithRegistersModel,Dinov2WithRegistersPreTrainedModel:()=>t.Dinov2WithRegistersPreTrainedModel,DistilBertForMaskedLM:()=>t.DistilBertForMaskedLM,DistilBertForQuestionAnswering:()=>t.DistilBertForQuestionAnswering,DistilBertForSequenceClassification:()=>t.DistilBertForSequenceClassification,DistilBertForTokenClassification:()=>t.DistilBertForTokenClassification,DistilBertModel:()=>t.DistilBertModel,DistilBertPreTrainedModel:()=>t.DistilBertPreTrainedModel,DistilBertTokenizer:()=>s.DistilBertTokenizer,DocumentQuestionAnsweringPipeline:()=>r.DocumentQuestionAnsweringPipeline,DonutFeatureExtractor:()=>f.DonutFeatureExtractor,DonutImageProcessor:()=>f.DonutImageProcessor,DonutSwinModel:()=>t.DonutSwinModel,DonutSwinPreTrainedModel:()=>t.DonutSwinPreTrainedModel,EfficientNetForImageClassification:()=>t.EfficientNetForImageClassification,EfficientNetImageProcessor:()=>f.EfficientNetImageProcessor,EfficientNetModel:()=>t.EfficientNetModel,EfficientNetPreTrainedModel:()=>t.EfficientNetPreTrainedModel,ElectraForMaskedLM:()=>t.ElectraForMaskedLM,ElectraForQuestionAnswering:()=>t.ElectraForQuestionAnswering,ElectraForSequenceClassification:()=>t.ElectraForSequenceClassification,ElectraForTokenClassification:()=>t.ElectraForTokenClassification,ElectraModel:()=>t.ElectraModel,ElectraPreTrainedModel:()=>t.ElectraPreTrainedModel,ElectraTokenizer:()=>s.ElectraTokenizer,EncodecFeatureExtractor:()=>d.EncodecFeatureExtractor,EosTokenCriteria:()=>x.EosTokenCriteria,Ernie4_5_ForCausalLM:()=>t.Ernie4_5_ForCausalLM,Ernie4_5_Model:()=>t.Ernie4_5_Model,Ernie4_5_PretrainedModel:()=>t.Ernie4_5_PretrainedModel,Ernie4_5_Tokenizer:()=>s.Ernie4_5_Tokenizer,EsmForMaskedLM:()=>t.EsmForMaskedLM,EsmForSequenceClassification:()=>t.EsmForSequenceClassification,EsmForTokenClassification:()=>t.EsmForTokenClassification,EsmModel:()=>t.EsmModel,EsmPreTrainedModel:()=>t.EsmPreTrainedModel,EsmTokenizer:()=>s.EsmTokenizer,ExaoneForCausalLM:()=>t.ExaoneForCausalLM,ExaoneModel:()=>t.ExaoneModel,ExaonePreTrainedModel:()=>t.ExaonePreTrainedModel,FFT:()=>c.FFT,FalconForCausalLM:()=>t.FalconForCausalLM,FalconModel:()=>t.FalconModel,FalconPreTrainedModel:()=>t.FalconPreTrainedModel,FalconTokenizer:()=>s.FalconTokenizer,FastViTForImageClassification:()=>t.FastViTForImageClassification,FastViTModel:()=>t.FastViTModel,FastViTPreTrainedModel:()=>t.FastViTPreTrainedModel,FeatureExtractionPipeline:()=>r.FeatureExtractionPipeline,FeatureExtractor:()=>p.FeatureExtractor,FillMaskPipeline:()=>r.FillMaskPipeline,Florence2ForConditionalGeneration:()=>t.Florence2ForConditionalGeneration,Florence2PreTrainedModel:()=>t.Florence2PreTrainedModel,Florence2Processor:()=>g.Florence2Processor,ForcedBOSTokenLogitsProcessor:()=>M.ForcedBOSTokenLogitsProcessor,ForcedEOSTokenLogitsProcessor:()=>M.ForcedEOSTokenLogitsProcessor,GLPNFeatureExtractor:()=>f.GLPNFeatureExtractor,GLPNForDepthEstimation:()=>t.GLPNForDepthEstimation,GLPNModel:()=>t.GLPNModel,GLPNPreTrainedModel:()=>t.GLPNPreTrainedModel,GPT2LMHeadModel:()=>t.GPT2LMHeadModel,GPT2Model:()=>t.GPT2Model,GPT2PreTrainedModel:()=>t.GPT2PreTrainedModel,GPT2Tokenizer:()=>s.GPT2Tokenizer,GPTBigCodeForCausalLM:()=>t.GPTBigCodeForCausalLM,GPTBigCodeModel:()=>t.GPTBigCodeModel,GPTBigCodePreTrainedModel:()=>t.GPTBigCodePreTrainedModel,GPTJForCausalLM:()=>t.GPTJForCausalLM,GPTJModel:()=>t.GPTJModel,GPTJPreTrainedModel:()=>t.GPTJPreTrainedModel,GPTNeoForCausalLM:()=>t.GPTNeoForCausalLM,GPTNeoModel:()=>t.GPTNeoModel,GPTNeoPreTrainedModel:()=>t.GPTNeoPreTrainedModel,GPTNeoXForCausalLM:()=>t.GPTNeoXForCausalLM,GPTNeoXModel:()=>t.GPTNeoXModel,GPTNeoXPreTrainedModel:()=>t.GPTNeoXPreTrainedModel,GPTNeoXTokenizer:()=>s.GPTNeoXTokenizer,Gemma2ForCausalLM:()=>t.Gemma2ForCausalLM,Gemma2Model:()=>t.Gemma2Model,Gemma2PreTrainedModel:()=>t.Gemma2PreTrainedModel,Gemma3ForCausalLM:()=>t.Gemma3ForCausalLM,Gemma3Model:()=>t.Gemma3Model,Gemma3PreTrainedModel:()=>t.Gemma3PreTrainedModel,Gemma3nAudioFeatureExtractor:()=>d.Gemma3nAudioFeatureExtractor,Gemma3nForConditionalGeneration:()=>t.Gemma3nForConditionalGeneration,Gemma3nPreTrainedModel:()=>t.Gemma3nPreTrainedModel,Gemma3nProcessor:()=>g.Gemma3nProcessor,GemmaForCausalLM:()=>t.GemmaForCausalLM,GemmaModel:()=>t.GemmaModel,GemmaPreTrainedModel:()=>t.GemmaPreTrainedModel,GemmaTokenizer:()=>s.GemmaTokenizer,GlmForCausalLM:()=>t.GlmForCausalLM,GlmModel:()=>t.GlmModel,GlmPreTrainedModel:()=>t.GlmPreTrainedModel,GraniteForCausalLM:()=>t.GraniteForCausalLM,GraniteModel:()=>t.GraniteModel,GraniteMoeHybridForCausalLM:()=>t.GraniteMoeHybridForCausalLM,GraniteMoeHybridModel:()=>t.GraniteMoeHybridModel,GraniteMoeHybridPreTrainedModel:()=>t.GraniteMoeHybridPreTrainedModel,GranitePreTrainedModel:()=>t.GranitePreTrainedModel,Grok1Tokenizer:()=>s.Grok1Tokenizer,GroundingDinoForObjectDetection:()=>t.GroundingDinoForObjectDetection,GroundingDinoImageProcessor:()=>f.GroundingDinoImageProcessor,GroundingDinoPreTrainedModel:()=>t.GroundingDinoPreTrainedModel,GroundingDinoProcessor:()=>g.GroundingDinoProcessor,GroupViTModel:()=>t.GroupViTModel,GroupViTPreTrainedModel:()=>t.GroupViTPreTrainedModel,HeliumForCausalLM:()=>t.HeliumForCausalLM,HeliumModel:()=>t.HeliumModel,HeliumPreTrainedModel:()=>t.HeliumPreTrainedModel,HerbertTokenizer:()=>s.HerbertTokenizer,HieraForImageClassification:()=>t.HieraForImageClassification,HieraModel:()=>t.HieraModel,HieraPreTrainedModel:()=>t.HieraPreTrainedModel,HubertForCTC:()=>t.HubertForCTC,HubertForSequenceClassification:()=>t.HubertForSequenceClassification,HubertModel:()=>t.HubertModel,HubertPreTrainedModel:()=>t.HubertPreTrainedModel,IJepaForImageClassification:()=>t.IJepaForImageClassification,IJepaModel:()=>t.IJepaModel,IJepaPreTrainedModel:()=>t.IJepaPreTrainedModel,Idefics3ForConditionalGeneration:()=>t.Idefics3ForConditionalGeneration,Idefics3ImageProcessor:()=>f.Idefics3ImageProcessor,Idefics3PreTrainedModel:()=>t.Idefics3PreTrainedModel,Idefics3Processor:()=>g.Idefics3Processor,ImageClassificationPipeline:()=>r.ImageClassificationPipeline,ImageFeatureExtractionPipeline:()=>r.ImageFeatureExtractionPipeline,ImageFeatureExtractor:()=>d.ImageFeatureExtractor,ImageMattingOutput:()=>t.ImageMattingOutput,ImageProcessor:()=>_.ImageProcessor,ImageSegmentationPipeline:()=>r.ImageSegmentationPipeline,ImageToImagePipeline:()=>r.ImageToImagePipeline,ImageToTextPipeline:()=>r.ImageToTextPipeline,InterruptableStoppingCriteria:()=>x.InterruptableStoppingCriteria,JAISLMHeadModel:()=>t.JAISLMHeadModel,JAISModel:()=>t.JAISModel,JAISPreTrainedModel:()=>t.JAISPreTrainedModel,JinaCLIPImageProcessor:()=>f.JinaCLIPImageProcessor,JinaCLIPModel:()=>t.JinaCLIPModel,JinaCLIPPreTrainedModel:()=>t.JinaCLIPPreTrainedModel,JinaCLIPProcessor:()=>g.JinaCLIPProcessor,JinaCLIPTextModel:()=>t.JinaCLIPTextModel,JinaCLIPVisionModel:()=>t.JinaCLIPVisionModel,Lfm2ForCausalLM:()=>t.Lfm2ForCausalLM,Lfm2Model:()=>t.Lfm2Model,Lfm2PreTrainedModel:()=>t.Lfm2PreTrainedModel,LiteWhisperForConditionalGeneration:()=>t.LiteWhisperForConditionalGeneration,Llama4ForCausalLM:()=>t.Llama4ForCausalLM,Llama4PreTrainedModel:()=>t.Llama4PreTrainedModel,LlamaForCausalLM:()=>t.LlamaForCausalLM,LlamaModel:()=>t.LlamaModel,LlamaPreTrainedModel:()=>t.LlamaPreTrainedModel,LlamaTokenizer:()=>s.LlamaTokenizer,LlavaForConditionalGeneration:()=>t.LlavaForConditionalGeneration,LlavaOnevisionForConditionalGeneration:()=>t.LlavaOnevisionForConditionalGeneration,LlavaOnevisionImageProcessor:()=>f.LlavaOnevisionImageProcessor,LlavaPreTrainedModel:()=>t.LlavaPreTrainedModel,LlavaProcessor:()=>g.LlavaProcessor,LlavaQwen2ForCausalLM:()=>t.LlavaQwen2ForCausalLM,LogitsProcessor:()=>M.LogitsProcessor,LogitsProcessorList:()=>M.LogitsProcessorList,LogitsWarper:()=>M.LogitsWarper,LongT5ForConditionalGeneration:()=>t.LongT5ForConditionalGeneration,LongT5Model:()=>t.LongT5Model,LongT5PreTrainedModel:()=>t.LongT5PreTrainedModel,M2M100ForConditionalGeneration:()=>t.M2M100ForConditionalGeneration,M2M100Model:()=>t.M2M100Model,M2M100PreTrainedModel:()=>t.M2M100PreTrainedModel,M2M100Tokenizer:()=>s.M2M100Tokenizer,MBart50Tokenizer:()=>s.MBart50Tokenizer,MBartForCausalLM:()=>t.MBartForCausalLM,MBartForConditionalGeneration:()=>t.MBartForConditionalGeneration,MBartForSequenceClassification:()=>t.MBartForSequenceClassification,MBartModel:()=>t.MBartModel,MBartPreTrainedModel:()=>t.MBartPreTrainedModel,MBartTokenizer:()=>s.MBartTokenizer,MPNetForMaskedLM:()=>t.MPNetForMaskedLM,MPNetForQuestionAnswering:()=>t.MPNetForQuestionAnswering,MPNetForSequenceClassification:()=>t.MPNetForSequenceClassification,MPNetForTokenClassification:()=>t.MPNetForTokenClassification,MPNetModel:()=>t.MPNetModel,MPNetPreTrainedModel:()=>t.MPNetPreTrainedModel,MPNetTokenizer:()=>s.MPNetTokenizer,MT5ForConditionalGeneration:()=>t.MT5ForConditionalGeneration,MT5Model:()=>t.MT5Model,MT5PreTrainedModel:()=>t.MT5PreTrainedModel,MarianMTModel:()=>t.MarianMTModel,MarianModel:()=>t.MarianModel,MarianPreTrainedModel:()=>t.MarianPreTrainedModel,MarianTokenizer:()=>s.MarianTokenizer,Mask2FormerImageProcessor:()=>f.Mask2FormerImageProcessor,MaskFormerFeatureExtractor:()=>f.MaskFormerFeatureExtractor,MaskFormerForInstanceSegmentation:()=>t.MaskFormerForInstanceSegmentation,MaskFormerImageProcessor:()=>f.MaskFormerImageProcessor,MaskFormerModel:()=>t.MaskFormerModel,MaskFormerPreTrainedModel:()=>t.MaskFormerPreTrainedModel,MaskedLMOutput:()=>t.MaskedLMOutput,MaxLengthCriteria:()=>x.MaxLengthCriteria,Metric3DForDepthEstimation:()=>t.Metric3DForDepthEstimation,Metric3DPreTrainedModel:()=>t.Metric3DPreTrainedModel,Metric3Dv2ForDepthEstimation:()=>t.Metric3Dv2ForDepthEstimation,Metric3Dv2PreTrainedModel:()=>t.Metric3Dv2PreTrainedModel,MgpstrForSceneTextRecognition:()=>t.MgpstrForSceneTextRecognition,MgpstrModelOutput:()=>t.MgpstrModelOutput,MgpstrPreTrainedModel:()=>t.MgpstrPreTrainedModel,MgpstrProcessor:()=>g.MgpstrProcessor,MgpstrTokenizer:()=>s.MgpstrTokenizer,MimiDecoderModel:()=>t.MimiDecoderModel,MimiDecoderOutput:()=>t.MimiDecoderOutput,MimiEncoderModel:()=>t.MimiEncoderModel,MimiEncoderOutput:()=>t.MimiEncoderOutput,MimiModel:()=>t.MimiModel,MimiPreTrainedModel:()=>t.MimiPreTrainedModel,MinLengthLogitsProcessor:()=>M.MinLengthLogitsProcessor,MinNewTokensLengthLogitsProcessor:()=>M.MinNewTokensLengthLogitsProcessor,MistralForCausalLM:()=>t.MistralForCausalLM,MistralModel:()=>t.MistralModel,MistralPreTrainedModel:()=>t.MistralPreTrainedModel,MobileBertForMaskedLM:()=>t.MobileBertForMaskedLM,MobileBertForQuestionAnswering:()=>t.MobileBertForQuestionAnswering,MobileBertForSequenceClassification:()=>t.MobileBertForSequenceClassification,MobileBertModel:()=>t.MobileBertModel,MobileBertPreTrainedModel:()=>t.MobileBertPreTrainedModel,MobileBertTokenizer:()=>s.MobileBertTokenizer,MobileLLMForCausalLM:()=>t.MobileLLMForCausalLM,MobileLLMModel:()=>t.MobileLLMModel,MobileLLMPreTrainedModel:()=>t.MobileLLMPreTrainedModel,MobileNetV1FeatureExtractor:()=>f.MobileNetV1FeatureExtractor,MobileNetV1ForImageClassification:()=>t.MobileNetV1ForImageClassification,MobileNetV1ForSemanticSegmentation:()=>t.MobileNetV1ForSemanticSegmentation,MobileNetV1ImageProcessor:()=>f.MobileNetV1ImageProcessor,MobileNetV1Model:()=>t.MobileNetV1Model,MobileNetV1PreTrainedModel:()=>t.MobileNetV1PreTrainedModel,MobileNetV2FeatureExtractor:()=>f.MobileNetV2FeatureExtractor,MobileNetV2ForImageClassification:()=>t.MobileNetV2ForImageClassification,MobileNetV2ForSemanticSegmentation:()=>t.MobileNetV2ForSemanticSegmentation,MobileNetV2ImageProcessor:()=>f.MobileNetV2ImageProcessor,MobileNetV2Model:()=>t.MobileNetV2Model,MobileNetV2PreTrainedModel:()=>t.MobileNetV2PreTrainedModel,MobileNetV3FeatureExtractor:()=>f.MobileNetV3FeatureExtractor,MobileNetV3ForImageClassification:()=>t.MobileNetV3ForImageClassification,MobileNetV3ForSemanticSegmentation:()=>t.MobileNetV3ForSemanticSegmentation,MobileNetV3ImageProcessor:()=>f.MobileNetV3ImageProcessor,MobileNetV3Model:()=>t.MobileNetV3Model,MobileNetV3PreTrainedModel:()=>t.MobileNetV3PreTrainedModel,MobileNetV4FeatureExtractor:()=>f.MobileNetV4FeatureExtractor,MobileNetV4ForImageClassification:()=>t.MobileNetV4ForImageClassification,MobileNetV4ForSemanticSegmentation:()=>t.MobileNetV4ForSemanticSegmentation,MobileNetV4ImageProcessor:()=>f.MobileNetV4ImageProcessor,MobileNetV4Model:()=>t.MobileNetV4Model,MobileNetV4PreTrainedModel:()=>t.MobileNetV4PreTrainedModel,MobileViTFeatureExtractor:()=>f.MobileViTFeatureExtractor,MobileViTForImageClassification:()=>t.MobileViTForImageClassification,MobileViTImageProcessor:()=>f.MobileViTImageProcessor,MobileViTModel:()=>t.MobileViTModel,MobileViTPreTrainedModel:()=>t.MobileViTPreTrainedModel,MobileViTV2ForImageClassification:()=>t.MobileViTV2ForImageClassification,MobileViTV2Model:()=>t.MobileViTV2Model,MobileViTV2PreTrainedModel:()=>t.MobileViTV2PreTrainedModel,ModelOutput:()=>t.ModelOutput,ModernBertDecoderForCausalLM:()=>t.ModernBertDecoderForCausalLM,ModernBertDecoderModel:()=>t.ModernBertDecoderModel,ModernBertDecoderPreTrainedModel:()=>t.ModernBertDecoderPreTrainedModel,ModernBertForMaskedLM:()=>t.ModernBertForMaskedLM,ModernBertForSequenceClassification:()=>t.ModernBertForSequenceClassification,ModernBertForTokenClassification:()=>t.ModernBertForTokenClassification,ModernBertModel:()=>t.ModernBertModel,ModernBertPreTrainedModel:()=>t.ModernBertPreTrainedModel,Moondream1ForConditionalGeneration:()=>t.Moondream1ForConditionalGeneration,MoonshineFeatureExtractor:()=>d.MoonshineFeatureExtractor,MoonshineForConditionalGeneration:()=>t.MoonshineForConditionalGeneration,MoonshineModel:()=>t.MoonshineModel,MoonshinePreTrainedModel:()=>t.MoonshinePreTrainedModel,MoonshineProcessor:()=>g.MoonshineProcessor,MptForCausalLM:()=>t.MptForCausalLM,MptModel:()=>t.MptModel,MptPreTrainedModel:()=>t.MptPreTrainedModel,MultiModalityCausalLM:()=>t.MultiModalityCausalLM,MultiModalityPreTrainedModel:()=>t.MultiModalityPreTrainedModel,MusicgenForCausalLM:()=>t.MusicgenForCausalLM,MusicgenForConditionalGeneration:()=>t.MusicgenForConditionalGeneration,MusicgenModel:()=>t.MusicgenModel,MusicgenPreTrainedModel:()=>t.MusicgenPreTrainedModel,NanoChatForCausalLM:()=>t.NanoChatForCausalLM,NanoChatModel:()=>t.NanoChatModel,NanoChatPreTrainedModel:()=>t.NanoChatPreTrainedModel,NeoBertForMaskedLM:()=>t.NeoBertForMaskedLM,NeoBertForQuestionAnswering:()=>t.NeoBertForQuestionAnswering,NeoBertForSequenceClassification:()=>t.NeoBertForSequenceClassification,NeoBertForTokenClassification:()=>t.NeoBertForTokenClassification,NeoBertModel:()=>t.NeoBertModel,NeoBertPreTrainedModel:()=>t.NeoBertPreTrainedModel,NllbTokenizer:()=>s.NllbTokenizer,NoBadWordsLogitsProcessor:()=>M.NoBadWordsLogitsProcessor,NoRepeatNGramLogitsProcessor:()=>M.NoRepeatNGramLogitsProcessor,NomicBertModel:()=>t.NomicBertModel,NomicBertPreTrainedModel:()=>t.NomicBertPreTrainedModel,NougatImageProcessor:()=>f.NougatImageProcessor,NougatTokenizer:()=>s.NougatTokenizer,OPTForCausalLM:()=>t.OPTForCausalLM,OPTModel:()=>t.OPTModel,OPTPreTrainedModel:()=>t.OPTPreTrainedModel,ObjectDetectionPipeline:()=>r.ObjectDetectionPipeline,Olmo2ForCausalLM:()=>t.Olmo2ForCausalLM,Olmo2Model:()=>t.Olmo2Model,Olmo2PreTrainedModel:()=>t.Olmo2PreTrainedModel,OlmoForCausalLM:()=>t.OlmoForCausalLM,OlmoModel:()=>t.OlmoModel,OlmoPreTrainedModel:()=>t.OlmoPreTrainedModel,OpenELMForCausalLM:()=>t.OpenELMForCausalLM,OpenELMModel:()=>t.OpenELMModel,OpenELMPreTrainedModel:()=>t.OpenELMPreTrainedModel,OwlViTFeatureExtractor:()=>f.OwlViTFeatureExtractor,OwlViTForObjectDetection:()=>t.OwlViTForObjectDetection,OwlViTImageProcessor:()=>f.OwlViTImageProcessor,OwlViTModel:()=>t.OwlViTModel,OwlViTPreTrainedModel:()=>t.OwlViTPreTrainedModel,OwlViTProcessor:()=>g.OwlViTProcessor,Owlv2ForObjectDetection:()=>t.Owlv2ForObjectDetection,Owlv2ImageProcessor:()=>f.Owlv2ImageProcessor,Owlv2Model:()=>t.Owlv2Model,Owlv2PreTrainedModel:()=>t.Owlv2PreTrainedModel,PaliGemmaForConditionalGeneration:()=>t.PaliGemmaForConditionalGeneration,PaliGemmaPreTrainedModel:()=>t.PaliGemmaPreTrainedModel,PaliGemmaProcessor:()=>g.PaliGemmaProcessor,ParakeetFeatureExtractor:()=>d.ParakeetFeatureExtractor,ParakeetForCTC:()=>t.ParakeetForCTC,ParakeetPreTrainedModel:()=>t.ParakeetPreTrainedModel,PatchTSMixerForPrediction:()=>t.PatchTSMixerForPrediction,PatchTSMixerModel:()=>t.PatchTSMixerModel,PatchTSMixerPreTrainedModel:()=>t.PatchTSMixerPreTrainedModel,PatchTSTForPrediction:()=>t.PatchTSTForPrediction,PatchTSTModel:()=>t.PatchTSTModel,PatchTSTPreTrainedModel:()=>t.PatchTSTPreTrainedModel,Phi3ForCausalLM:()=>t.Phi3ForCausalLM,Phi3Model:()=>t.Phi3Model,Phi3PreTrainedModel:()=>t.Phi3PreTrainedModel,Phi3VForCausalLM:()=>t.Phi3VForCausalLM,Phi3VImageProcessor:()=>f.Phi3VImageProcessor,Phi3VPreTrainedModel:()=>t.Phi3VPreTrainedModel,Phi3VProcessor:()=>g.Phi3VProcessor,PhiForCausalLM:()=>t.PhiForCausalLM,PhiModel:()=>t.PhiModel,PhiPreTrainedModel:()=>t.PhiPreTrainedModel,Pipeline:()=>r.Pipeline,PreTrainedModel:()=>t.PreTrainedModel,PreTrainedTokenizer:()=>s.PreTrainedTokenizer,PretrainedConfig:()=>o.PretrainedConfig,PretrainedMixin:()=>t.PretrainedMixin,Processor:()=>A.Processor,PvtForImageClassification:()=>t.PvtForImageClassification,PvtImageProcessor:()=>f.PvtImageProcessor,PvtModel:()=>t.PvtModel,PvtPreTrainedModel:()=>t.PvtPreTrainedModel,PyAnnoteFeatureExtractor:()=>d.PyAnnoteFeatureExtractor,PyAnnoteForAudioFrameClassification:()=>t.PyAnnoteForAudioFrameClassification,PyAnnoteModel:()=>t.PyAnnoteModel,PyAnnotePreTrainedModel:()=>t.PyAnnotePreTrainedModel,PyAnnoteProcessor:()=>g.PyAnnoteProcessor,QuestionAnsweringModelOutput:()=>t.QuestionAnsweringModelOutput,QuestionAnsweringPipeline:()=>r.QuestionAnsweringPipeline,Qwen2ForCausalLM:()=>t.Qwen2ForCausalLM,Qwen2Model:()=>t.Qwen2Model,Qwen2PreTrainedModel:()=>t.Qwen2PreTrainedModel,Qwen2Tokenizer:()=>s.Qwen2Tokenizer,Qwen2VLForConditionalGeneration:()=>t.Qwen2VLForConditionalGeneration,Qwen2VLImageProcessor:()=>f.Qwen2VLImageProcessor,Qwen2VLPreTrainedModel:()=>t.Qwen2VLPreTrainedModel,Qwen2VLProcessor:()=>g.Qwen2VLProcessor,Qwen3ForCausalLM:()=>t.Qwen3ForCausalLM,Qwen3Model:()=>t.Qwen3Model,Qwen3PreTrainedModel:()=>t.Qwen3PreTrainedModel,RFDetrForObjectDetection:()=>t.RFDetrForObjectDetection,RFDetrModel:()=>t.RFDetrModel,RFDetrObjectDetectionOutput:()=>t.RFDetrObjectDetectionOutput,RFDetrPreTrainedModel:()=>t.RFDetrPreTrainedModel,RTDetrForObjectDetection:()=>t.RTDetrForObjectDetection,RTDetrImageProcessor:()=>f.RTDetrImageProcessor,RTDetrModel:()=>t.RTDetrModel,RTDetrObjectDetectionOutput:()=>t.RTDetrObjectDetectionOutput,RTDetrPreTrainedModel:()=>t.RTDetrPreTrainedModel,RTDetrV2ForObjectDetection:()=>t.RTDetrV2ForObjectDetection,RTDetrV2Model:()=>t.RTDetrV2Model,RTDetrV2ObjectDetectionOutput:()=>t.RTDetrV2ObjectDetectionOutput,RTDetrV2PreTrainedModel:()=>t.RTDetrV2PreTrainedModel,RawAudio:()=>n.RawAudio,RawImage:()=>a.RawImage,RawVideo:()=>i.RawVideo,RawVideoFrame:()=>i.RawVideoFrame,RepetitionPenaltyLogitsProcessor:()=>M.RepetitionPenaltyLogitsProcessor,ResNetForImageClassification:()=>t.ResNetForImageClassification,ResNetModel:()=>t.ResNetModel,ResNetPreTrainedModel:()=>t.ResNetPreTrainedModel,RoFormerForMaskedLM:()=>t.RoFormerForMaskedLM,RoFormerForQuestionAnswering:()=>t.RoFormerForQuestionAnswering,RoFormerForSequenceClassification:()=>t.RoFormerForSequenceClassification,RoFormerForTokenClassification:()=>t.RoFormerForTokenClassification,RoFormerModel:()=>t.RoFormerModel,RoFormerPreTrainedModel:()=>t.RoFormerPreTrainedModel,RoFormerTokenizer:()=>s.RoFormerTokenizer,RobertaForMaskedLM:()=>t.RobertaForMaskedLM,RobertaForQuestionAnswering:()=>t.RobertaForQuestionAnswering,RobertaForSequenceClassification:()=>t.RobertaForSequenceClassification,RobertaForTokenClassification:()=>t.RobertaForTokenClassification,RobertaModel:()=>t.RobertaModel,RobertaPreTrainedModel:()=>t.RobertaPreTrainedModel,RobertaTokenizer:()=>s.RobertaTokenizer,SamImageProcessor:()=>f.SamImageProcessor,SamImageSegmentationOutput:()=>t.SamImageSegmentationOutput,SamModel:()=>t.SamModel,SamPreTrainedModel:()=>t.SamPreTrainedModel,SamProcessor:()=>g.SamProcessor,SapiensForDepthEstimation:()=>t.SapiensForDepthEstimation,SapiensForNormalEstimation:()=>t.SapiensForNormalEstimation,SapiensForSemanticSegmentation:()=>t.SapiensForSemanticSegmentation,SapiensPreTrainedModel:()=>t.SapiensPreTrainedModel,SeamlessM4TFeatureExtractor:()=>d.SeamlessM4TFeatureExtractor,SegformerFeatureExtractor:()=>f.SegformerFeatureExtractor,SegformerForImageClassification:()=>t.SegformerForImageClassification,SegformerForSemanticSegmentation:()=>t.SegformerForSemanticSegmentation,SegformerImageProcessor:()=>f.SegformerImageProcessor,SegformerModel:()=>t.SegformerModel,SegformerPreTrainedModel:()=>t.SegformerPreTrainedModel,Seq2SeqLMOutput:()=>t.Seq2SeqLMOutput,SequenceClassifierOutput:()=>t.SequenceClassifierOutput,SiglipImageProcessor:()=>f.SiglipImageProcessor,SiglipModel:()=>t.SiglipModel,SiglipPreTrainedModel:()=>t.SiglipPreTrainedModel,SiglipTextModel:()=>t.SiglipTextModel,SiglipTokenizer:()=>s.SiglipTokenizer,SiglipVisionModel:()=>t.SiglipVisionModel,SmolLM3ForCausalLM:()=>t.SmolLM3ForCausalLM,SmolLM3Model:()=>t.SmolLM3Model,SmolLM3PreTrainedModel:()=>t.SmolLM3PreTrainedModel,SmolVLMForConditionalGeneration:()=>t.SmolVLMForConditionalGeneration,SmolVLMImageProcessor:()=>f.SmolVLMImageProcessor,SmolVLMProcessor:()=>g.SmolVLMProcessor,SnacDecoderModel:()=>t.SnacDecoderModel,SnacEncoderModel:()=>t.SnacEncoderModel,SnacFeatureExtractor:()=>d.SnacFeatureExtractor,SnacModel:()=>t.SnacModel,SnacPreTrainedModel:()=>t.SnacPreTrainedModel,SpeechT5FeatureExtractor:()=>d.SpeechT5FeatureExtractor,SpeechT5ForSpeechToText:()=>t.SpeechT5ForSpeechToText,SpeechT5ForTextToSpeech:()=>t.SpeechT5ForTextToSpeech,SpeechT5HifiGan:()=>t.SpeechT5HifiGan,SpeechT5Model:()=>t.SpeechT5Model,SpeechT5PreTrainedModel:()=>t.SpeechT5PreTrainedModel,SpeechT5Processor:()=>g.SpeechT5Processor,SpeechT5Tokenizer:()=>s.SpeechT5Tokenizer,SqueezeBertForMaskedLM:()=>t.SqueezeBertForMaskedLM,SqueezeBertForQuestionAnswering:()=>t.SqueezeBertForQuestionAnswering,SqueezeBertForSequenceClassification:()=>t.SqueezeBertForSequenceClassification,SqueezeBertModel:()=>t.SqueezeBertModel,SqueezeBertPreTrainedModel:()=>t.SqueezeBertPreTrainedModel,SqueezeBertTokenizer:()=>s.SqueezeBertTokenizer,StableLmForCausalLM:()=>t.StableLmForCausalLM,StableLmModel:()=>t.StableLmModel,StableLmPreTrainedModel:()=>t.StableLmPreTrainedModel,Starcoder2ForCausalLM:()=>t.Starcoder2ForCausalLM,Starcoder2Model:()=>t.Starcoder2Model,Starcoder2PreTrainedModel:()=>t.Starcoder2PreTrainedModel,StoppingCriteria:()=>x.StoppingCriteria,StoppingCriteriaList:()=>x.StoppingCriteriaList,StyleTextToSpeech2Model:()=>t.StyleTextToSpeech2Model,StyleTextToSpeech2PreTrainedModel:()=>t.StyleTextToSpeech2PreTrainedModel,SummarizationPipeline:()=>r.SummarizationPipeline,SuppressTokensAtBeginLogitsProcessor:()=>M.SuppressTokensAtBeginLogitsProcessor,Swin2SRForImageSuperResolution:()=>t.Swin2SRForImageSuperResolution,Swin2SRImageProcessor:()=>f.Swin2SRImageProcessor,Swin2SRModel:()=>t.Swin2SRModel,Swin2SRPreTrainedModel:()=>t.Swin2SRPreTrainedModel,SwinForImageClassification:()=>t.SwinForImageClassification,SwinForSemanticSegmentation:()=>t.SwinForSemanticSegmentation,SwinModel:()=>t.SwinModel,SwinPreTrainedModel:()=>t.SwinPreTrainedModel,T5ForConditionalGeneration:()=>t.T5ForConditionalGeneration,T5Model:()=>t.T5Model,T5PreTrainedModel:()=>t.T5PreTrainedModel,T5Tokenizer:()=>s.T5Tokenizer,TableTransformerForObjectDetection:()=>t.TableTransformerForObjectDetection,TableTransformerModel:()=>t.TableTransformerModel,TableTransformerObjectDetectionOutput:()=>t.TableTransformerObjectDetectionOutput,TableTransformerPreTrainedModel:()=>t.TableTransformerPreTrainedModel,TemperatureLogitsWarper:()=>M.TemperatureLogitsWarper,Tensor:()=>l.Tensor,Text2TextGenerationPipeline:()=>r.Text2TextGenerationPipeline,TextClassificationPipeline:()=>r.TextClassificationPipeline,TextGenerationPipeline:()=>r.TextGenerationPipeline,TextStreamer:()=>C.TextStreamer,TextToAudioPipeline:()=>r.TextToAudioPipeline,TokenClassificationPipeline:()=>r.TokenClassificationPipeline,TokenClassifierOutput:()=>t.TokenClassifierOutput,TokenizerModel:()=>s.TokenizerModel,TopKLogitsWarper:()=>M.TopKLogitsWarper,TopPLogitsWarper:()=>M.TopPLogitsWarper,TrOCRForCausalLM:()=>t.TrOCRForCausalLM,TrOCRPreTrainedModel:()=>t.TrOCRPreTrainedModel,TranslationPipeline:()=>r.TranslationPipeline,UltravoxModel:()=>t.UltravoxModel,UltravoxPreTrainedModel:()=>t.UltravoxPreTrainedModel,UltravoxProcessor:()=>g.UltravoxProcessor,UniSpeechForCTC:()=>t.UniSpeechForCTC,UniSpeechForSequenceClassification:()=>t.UniSpeechForSequenceClassification,UniSpeechModel:()=>t.UniSpeechModel,UniSpeechPreTrainedModel:()=>t.UniSpeechPreTrainedModel,UniSpeechSatForAudioFrameClassification:()=>t.UniSpeechSatForAudioFrameClassification,UniSpeechSatForCTC:()=>t.UniSpeechSatForCTC,UniSpeechSatForSequenceClassification:()=>t.UniSpeechSatForSequenceClassification,UniSpeechSatModel:()=>t.UniSpeechSatModel,UniSpeechSatPreTrainedModel:()=>t.UniSpeechSatPreTrainedModel,VLChatProcessor:()=>g.VLChatProcessor,VLMImageProcessor:()=>f.VLMImageProcessor,VaultGemmaForCausalLM:()=>t.VaultGemmaForCausalLM,VaultGemmaModel:()=>t.VaultGemmaModel,VaultGemmaPreTrainedModel:()=>t.VaultGemmaPreTrainedModel,ViTFeatureExtractor:()=>f.ViTFeatureExtractor,ViTForImageClassification:()=>t.ViTForImageClassification,ViTImageProcessor:()=>f.ViTImageProcessor,ViTMAEModel:()=>t.ViTMAEModel,ViTMAEPreTrainedModel:()=>t.ViTMAEPreTrainedModel,ViTMSNForImageClassification:()=>t.ViTMSNForImageClassification,ViTMSNModel:()=>t.ViTMSNModel,ViTMSNPreTrainedModel:()=>t.ViTMSNPreTrainedModel,ViTModel:()=>t.ViTModel,ViTPreTrainedModel:()=>t.ViTPreTrainedModel,VisionEncoderDecoderModel:()=>t.VisionEncoderDecoderModel,VitMatteForImageMatting:()=>t.VitMatteForImageMatting,VitMatteImageProcessor:()=>f.VitMatteImageProcessor,VitMattePreTrainedModel:()=>t.VitMattePreTrainedModel,VitPoseForPoseEstimation:()=>t.VitPoseForPoseEstimation,VitPoseImageProcessor:()=>f.VitPoseImageProcessor,VitPosePreTrainedModel:()=>t.VitPosePreTrainedModel,VitsModel:()=>t.VitsModel,VitsModelOutput:()=>t.VitsModelOutput,VitsPreTrainedModel:()=>t.VitsPreTrainedModel,VitsTokenizer:()=>s.VitsTokenizer,VoxtralForConditionalGeneration:()=>t.VoxtralForConditionalGeneration,VoxtralProcessor:()=>g.VoxtralProcessor,Wav2Vec2BertForCTC:()=>t.Wav2Vec2BertForCTC,Wav2Vec2BertForSequenceClassification:()=>t.Wav2Vec2BertForSequenceClassification,Wav2Vec2BertModel:()=>t.Wav2Vec2BertModel,Wav2Vec2BertPreTrainedModel:()=>t.Wav2Vec2BertPreTrainedModel,Wav2Vec2CTCTokenizer:()=>s.Wav2Vec2CTCTokenizer,Wav2Vec2FeatureExtractor:()=>d.Wav2Vec2FeatureExtractor,Wav2Vec2ForAudioFrameClassification:()=>t.Wav2Vec2ForAudioFrameClassification,Wav2Vec2ForCTC:()=>t.Wav2Vec2ForCTC,Wav2Vec2ForSequenceClassification:()=>t.Wav2Vec2ForSequenceClassification,Wav2Vec2Model:()=>t.Wav2Vec2Model,Wav2Vec2PreTrainedModel:()=>t.Wav2Vec2PreTrainedModel,Wav2Vec2Processor:()=>g.Wav2Vec2Processor,Wav2Vec2ProcessorWithLM:()=>g.Wav2Vec2ProcessorWithLM,WavLMForAudioFrameClassification:()=>t.WavLMForAudioFrameClassification,WavLMForCTC:()=>t.WavLMForCTC,WavLMForSequenceClassification:()=>t.WavLMForSequenceClassification,WavLMForXVector:()=>t.WavLMForXVector,WavLMModel:()=>t.WavLMModel,WavLMPreTrainedModel:()=>t.WavLMPreTrainedModel,WeSpeakerFeatureExtractor:()=>d.WeSpeakerFeatureExtractor,WeSpeakerResNetModel:()=>t.WeSpeakerResNetModel,WeSpeakerResNetPreTrainedModel:()=>t.WeSpeakerResNetPreTrainedModel,WhisperFeatureExtractor:()=>d.WhisperFeatureExtractor,WhisperForConditionalGeneration:()=>t.WhisperForConditionalGeneration,WhisperModel:()=>t.WhisperModel,WhisperPreTrainedModel:()=>t.WhisperPreTrainedModel,WhisperProcessor:()=>g.WhisperProcessor,WhisperTextStreamer:()=>C.WhisperTextStreamer,WhisperTimeStampLogitsProcessor:()=>M.WhisperTimeStampLogitsProcessor,WhisperTokenizer:()=>s.WhisperTokenizer,XLMForQuestionAnswering:()=>t.XLMForQuestionAnswering,XLMForSequenceClassification:()=>t.XLMForSequenceClassification,XLMForTokenClassification:()=>t.XLMForTokenClassification,XLMModel:()=>t.XLMModel,XLMPreTrainedModel:()=>t.XLMPreTrainedModel,XLMRobertaForMaskedLM:()=>t.XLMRobertaForMaskedLM,XLMRobertaForQuestionAnswering:()=>t.XLMRobertaForQuestionAnswering,XLMRobertaForSequenceClassification:()=>t.XLMRobertaForSequenceClassification,XLMRobertaForTokenClassification:()=>t.XLMRobertaForTokenClassification,XLMRobertaModel:()=>t.XLMRobertaModel,XLMRobertaPreTrainedModel:()=>t.XLMRobertaPreTrainedModel,XLMRobertaTokenizer:()=>s.XLMRobertaTokenizer,XLMTokenizer:()=>s.XLMTokenizer,XLMWithLMHeadModel:()=>t.XLMWithLMHeadModel,XVectorOutput:()=>t.XVectorOutput,YolosFeatureExtractor:()=>f.YolosFeatureExtractor,YolosForObjectDetection:()=>t.YolosForObjectDetection,YolosImageProcessor:()=>f.YolosImageProcessor,YolosModel:()=>t.YolosModel,YolosObjectDetectionOutput:()=>t.YolosObjectDetectionOutput,YolosPreTrainedModel:()=>t.YolosPreTrainedModel,ZeroShotAudioClassificationPipeline:()=>r.ZeroShotAudioClassificationPipeline,ZeroShotClassificationPipeline:()=>r.ZeroShotClassificationPipeline,ZeroShotImageClassificationPipeline:()=>r.ZeroShotImageClassificationPipeline,ZeroShotObjectDetectionPipeline:()=>r.ZeroShotObjectDetectionPipeline,bankers_round:()=>c.bankers_round,cat:()=>l.cat,cos_sim:()=>c.cos_sim,dot:()=>c.dot,dynamic_time_warping:()=>c.dynamic_time_warping,env:()=>e.env,full:()=>l.full,full_like:()=>l.full_like,getCacheShapes:()=>o.getCacheShapes,hamming:()=>n.hamming,hanning:()=>n.hanning,interpolate:()=>l.interpolate,interpolate_4d:()=>l.interpolate_4d,interpolate_data:()=>c.interpolate_data,is_chinese_char:()=>s.is_chinese_char,layer_norm:()=>l.layer_norm,load_image:()=>a.load_image,load_video:()=>i.load_video,log_softmax:()=>c.log_softmax,magnitude:()=>c.magnitude,matmul:()=>l.matmul,max:()=>c.max,mean:()=>l.mean,mean_pooling:()=>l.mean_pooling,medianFilter:()=>c.medianFilter,mel_filter_bank:()=>n.mel_filter_bank,min:()=>c.min,ones:()=>l.ones,ones_like:()=>l.ones_like,permute:()=>l.permute,permute_data:()=>c.permute_data,pipeline:()=>r.pipeline,quantize_embeddings:()=>l.quantize_embeddings,rand:()=>l.rand,read_audio:()=>n.read_audio,rfft:()=>l.rfft,round:()=>c.round,slice:()=>l.slice,softmax:()=>c.softmax,spectrogram:()=>n.spectrogram,stack:()=>l.stack,std_mean:()=>l.std_mean,topk:()=>l.topk,window_function:()=>n.window_function,zeros:()=>l.zeros,zeros_like:()=>l.zeros_like});var e=Vt("./src/env.js"),r=Vt("./src/pipelines.js"),t=Vt("./src/models.js"),s=Vt("./src/tokenizers.js"),o=Vt("./src/configs.js"),n=Vt("./src/utils/audio.js"),a=Vt("./src/utils/image.js"),i=Vt("./src/utils/video.js"),l=Vt("./src/utils/tensor.js"),c=Vt("./src/utils/maths.js"),p=Vt("./src/base/feature_extraction_utils.js"),d=Vt("./src/models/feature_extractors.js"),u=Vt("./src/models/auto/feature_extraction_auto.js"),_=Vt("./src/base/image_processors_utils.js"),f=Vt("./src/models/image_processors.js"),b=Vt("./src/models/auto/image_processing_auto.js"),A=Vt("./src/base/processing_utils.js"),g=Vt("./src/models/processors.js"),y=Vt("./src/models/auto/processing_auto.js"),C=Vt("./src/generation/streamers.js"),x=Vt("./src/generation/stopping_criteria.js"),M=Vt("./src/generation/logits_process.js")})(),m.ASTFeatureExtractor,m.ASTForAudioClassification,m.ASTModel,m.ASTPreTrainedModel,m.AlbertForMaskedLM,m.AlbertForQuestionAnswering,m.AlbertForSequenceClassification,m.AlbertModel,m.AlbertPreTrainedModel,m.AlbertTokenizer,m.ArceeForCausalLM,m.ArceeModel,m.ArceePreTrainedModel,m.AudioClassificationPipeline,m.AutoConfig,m.AutoFeatureExtractor,m.AutoImageProcessor,m.AutoModel,m.AutoModelForAudioClassification,m.AutoModelForAudioFrameClassification,m.AutoModelForAudioTextToText,m.AutoModelForCTC,m.AutoModelForCausalLM,m.AutoModelForDepthEstimation,m.AutoModelForDocumentQuestionAnswering,m.AutoModelForImageClassification,m.AutoModelForImageFeatureExtraction,m.AutoModelForImageMatting,m.AutoModelForImageSegmentation;var GT=m.AutoModelForImageTextToText;m.AutoModelForImageToImage,m.AutoModelForMaskGeneration,m.AutoModelForMaskedLM,m.AutoModelForNormalEstimation,m.AutoModelForObjectDetection,m.AutoModelForPoseEstimation,m.AutoModelForQuestionAnswering,m.AutoModelForSemanticSegmentation,m.AutoModelForSeq2SeqLM,m.AutoModelForSequenceClassification,m.AutoModelForSpeechSeq2Seq,m.AutoModelForTextToSpectrogram,m.AutoModelForTextToWaveform,m.AutoModelForTokenClassification,m.AutoModelForUniversalSegmentation,m.AutoModelForVision2Seq,m.AutoModelForXVector,m.AutoModelForZeroShotObjectDetection;var HT=m.AutoProcessor;m.AutoTokenizer,m.AutomaticSpeechRecognitionPipeline,m.BackgroundRemovalPipeline,m.BartForConditionalGeneration,m.BartForSequenceClassification,m.BartModel,m.BartPretrainedModel,m.BartTokenizer,m.BaseModelOutput,m.BaseStreamer,m.BeitFeatureExtractor,m.BeitForImageClassification,m.BeitModel,m.BeitPreTrainedModel,m.BertForMaskedLM,m.BertForQuestionAnswering,m.BertForSequenceClassification,m.BertForTokenClassification,m.BertModel,m.BertPreTrainedModel,m.BertTokenizer,m.BitImageProcessor,m.BlenderbotForConditionalGeneration,m.BlenderbotModel,m.BlenderbotPreTrainedModel,m.BlenderbotSmallForConditionalGeneration,m.BlenderbotSmallModel,m.BlenderbotSmallPreTrainedModel,m.BlenderbotSmallTokenizer,m.BlenderbotTokenizer,m.BloomForCausalLM,m.BloomModel,m.BloomPreTrainedModel,m.BloomTokenizer,m.CLIPFeatureExtractor,m.CLIPImageProcessor,m.CLIPModel,m.CLIPPreTrainedModel,m.CLIPSegForImageSegmentation,m.CLIPSegModel,m.CLIPSegPreTrainedModel,m.CLIPTextModel,m.CLIPTextModelWithProjection,m.CLIPTokenizer,m.CLIPVisionModel,m.CLIPVisionModelWithProjection,m.CamembertForMaskedLM,m.CamembertForQuestionAnswering,m.CamembertForSequenceClassification,m.CamembertForTokenClassification,m.CamembertModel,m.CamembertPreTrainedModel,m.CamembertTokenizer,m.CausalLMOutput,m.CausalLMOutputWithPast,m.ChineseCLIPFeatureExtractor,m.ChineseCLIPModel,m.ChineseCLIPPreTrainedModel,m.ClapAudioModelWithProjection,m.ClapFeatureExtractor,m.ClapModel,m.ClapPreTrainedModel,m.ClapTextModelWithProjection,m.ClassifierFreeGuidanceLogitsProcessor,m.CodeGenForCausalLM,m.CodeGenModel,m.CodeGenPreTrainedModel,m.CodeGenTokenizer,m.CodeLlamaTokenizer,m.CohereForCausalLM,m.CohereModel,m.CoherePreTrainedModel,m.CohereTokenizer,m.ConvBertForMaskedLM,m.ConvBertForQuestionAnswering,m.ConvBertForSequenceClassification,m.ConvBertForTokenClassification,m.ConvBertModel,m.ConvBertPreTrainedModel,m.ConvBertTokenizer,m.ConvNextFeatureExtractor,m.ConvNextForImageClassification,m.ConvNextImageProcessor,m.ConvNextModel,m.ConvNextPreTrainedModel,m.ConvNextV2ForImageClassification,m.ConvNextV2Model,m.ConvNextV2PreTrainedModel,m.DFineForObjectDetection,m.DFineModel,m.DFinePreTrainedModel,m.DINOv3ConvNextModel,m.DINOv3ConvNextPreTrainedModel,m.DINOv3ViTImageProcessor,m.DINOv3ViTModel,m.DINOv3ViTPreTrainedModel,m.DPTFeatureExtractor,m.DPTForDepthEstimation,m.DPTImageProcessor,m.DPTModel,m.DPTPreTrainedModel,m.DacDecoderModel,m.DacDecoderOutput,m.DacEncoderModel,m.DacEncoderOutput,m.DacFeatureExtractor,m.DacModel,m.DacPreTrainedModel,m.DataTypeMap,m.DebertaForMaskedLM,m.DebertaForQuestionAnswering,m.DebertaForSequenceClassification,m.DebertaForTokenClassification,m.DebertaModel,m.DebertaPreTrainedModel,m.DebertaTokenizer,m.DebertaV2ForMaskedLM,m.DebertaV2ForQuestionAnswering,m.DebertaV2ForSequenceClassification,m.DebertaV2ForTokenClassification,m.DebertaV2Model,m.DebertaV2PreTrainedModel,m.DebertaV2Tokenizer,m.DecisionTransformerModel,m.DecisionTransformerPreTrainedModel,m.DeiTFeatureExtractor,m.DeiTForImageClassification,m.DeiTImageProcessor,m.DeiTModel,m.DeiTPreTrainedModel,m.DepthAnythingForDepthEstimation,m.DepthAnythingPreTrainedModel,m.DepthEstimationPipeline,m.DepthProForDepthEstimation,m.DepthProPreTrainedModel,m.DetrFeatureExtractor,m.DetrForObjectDetection,m.DetrForSegmentation,m.DetrImageProcessor,m.DetrModel,m.DetrObjectDetectionOutput,m.DetrPreTrainedModel,m.DetrSegmentationOutput,m.Dinov2ForImageClassification,m.Dinov2Model,m.Dinov2PreTrainedModel,m.Dinov2WithRegistersForImageClassification,m.Dinov2WithRegistersModel,m.Dinov2WithRegistersPreTrainedModel,m.DistilBertForMaskedLM,m.DistilBertForQuestionAnswering,m.DistilBertForSequenceClassification,m.DistilBertForTokenClassification,m.DistilBertModel,m.DistilBertPreTrainedModel,m.DistilBertTokenizer,m.DocumentQuestionAnsweringPipeline,m.DonutFeatureExtractor,m.DonutImageProcessor,m.DonutSwinModel,m.DonutSwinPreTrainedModel,m.EfficientNetForImageClassification,m.EfficientNetImageProcessor,m.EfficientNetModel,m.EfficientNetPreTrainedModel,m.ElectraForMaskedLM,m.ElectraForQuestionAnswering,m.ElectraForSequenceClassification,m.ElectraForTokenClassification,m.ElectraModel,m.ElectraPreTrainedModel,m.ElectraTokenizer,m.EncodecFeatureExtractor,m.EosTokenCriteria,m.Ernie4_5_ForCausalLM,m.Ernie4_5_Model,m.Ernie4_5_PretrainedModel,m.Ernie4_5_Tokenizer,m.EsmForMaskedLM,m.EsmForSequenceClassification,m.EsmForTokenClassification,m.EsmModel,m.EsmPreTrainedModel,m.EsmTokenizer,m.ExaoneForCausalLM,m.ExaoneModel,m.ExaonePreTrainedModel,m.FFT,m.FalconForCausalLM,m.FalconModel,m.FalconPreTrainedModel,m.FalconTokenizer,m.FastViTForImageClassification,m.FastViTModel,m.FastViTPreTrainedModel,m.FeatureExtractionPipeline,m.FeatureExtractor,m.FillMaskPipeline,m.Florence2ForConditionalGeneration,m.Florence2PreTrainedModel,m.Florence2Processor,m.ForcedBOSTokenLogitsProcessor,m.ForcedEOSTokenLogitsProcessor,m.GLPNFeatureExtractor,m.GLPNForDepthEstimation,m.GLPNModel,m.GLPNPreTrainedModel,m.GPT2LMHeadModel,m.GPT2Model,m.GPT2PreTrainedModel,m.GPT2Tokenizer,m.GPTBigCodeForCausalLM,m.GPTBigCodeModel,m.GPTBigCodePreTrainedModel,m.GPTJForCausalLM,m.GPTJModel,m.GPTJPreTrainedModel,m.GPTNeoForCausalLM,m.GPTNeoModel,m.GPTNeoPreTrainedModel,m.GPTNeoXForCausalLM,m.GPTNeoXModel,m.GPTNeoXPreTrainedModel,m.GPTNeoXTokenizer,m.Gemma2ForCausalLM,m.Gemma2Model,m.Gemma2PreTrainedModel,m.Gemma3ForCausalLM,m.Gemma3Model,m.Gemma3PreTrainedModel,m.Gemma3nAudioFeatureExtractor,m.Gemma3nForConditionalGeneration,m.Gemma3nPreTrainedModel,m.Gemma3nProcessor,m.GemmaForCausalLM,m.GemmaModel,m.GemmaPreTrainedModel,m.GemmaTokenizer,m.GlmForCausalLM,m.GlmModel,m.GlmPreTrainedModel,m.GraniteForCausalLM,m.GraniteModel,m.GraniteMoeHybridForCausalLM,m.GraniteMoeHybridModel,m.GraniteMoeHybridPreTrainedModel,m.GranitePreTrainedModel,m.Grok1Tokenizer,m.GroundingDinoForObjectDetection,m.GroundingDinoImageProcessor,m.GroundingDinoPreTrainedModel,m.GroundingDinoProcessor,m.GroupViTModel,m.GroupViTPreTrainedModel,m.HeliumForCausalLM,m.HeliumModel,m.HeliumPreTrainedModel,m.HerbertTokenizer,m.HieraForImageClassification,m.HieraModel,m.HieraPreTrainedModel,m.HubertForCTC,m.HubertForSequenceClassification,m.HubertModel,m.HubertPreTrainedModel,m.IJepaForImageClassification,m.IJepaModel,m.IJepaPreTrainedModel,m.Idefics3ForConditionalGeneration,m.Idefics3ImageProcessor,m.Idefics3PreTrainedModel,m.Idefics3Processor,m.ImageClassificationPipeline,m.ImageFeatureExtractionPipeline,m.ImageFeatureExtractor,m.ImageMattingOutput,m.ImageProcessor,m.ImageSegmentationPipeline,m.ImageToImagePipeline,m.ImageToTextPipeline,m.InterruptableStoppingCriteria,m.JAISLMHeadModel,m.JAISModel,m.JAISPreTrainedModel,m.JinaCLIPImageProcessor,m.JinaCLIPModel,m.JinaCLIPPreTrainedModel,m.JinaCLIPProcessor,m.JinaCLIPTextModel,m.JinaCLIPVisionModel,m.Lfm2ForCausalLM,m.Lfm2Model,m.Lfm2PreTrainedModel,m.LiteWhisperForConditionalGeneration,m.Llama4ForCausalLM,m.Llama4PreTrainedModel,m.LlamaForCausalLM,m.LlamaModel,m.LlamaPreTrainedModel,m.LlamaTokenizer,m.LlavaForConditionalGeneration,m.LlavaOnevisionForConditionalGeneration,m.LlavaOnevisionImageProcessor,m.LlavaPreTrainedModel,m.LlavaProcessor,m.LlavaQwen2ForCausalLM,m.LogitsProcessor,m.LogitsProcessorList,m.LogitsWarper,m.LongT5ForConditionalGeneration,m.LongT5Model,m.LongT5PreTrainedModel,m.M2M100ForConditionalGeneration,m.M2M100Model,m.M2M100PreTrainedModel,m.M2M100Tokenizer,m.MBart50Tokenizer,m.MBartForCausalLM,m.MBartForConditionalGeneration,m.MBartForSequenceClassification,m.MBartModel,m.MBartPreTrainedModel,m.MBartTokenizer,m.MPNetForMaskedLM,m.MPNetForQuestionAnswering,m.MPNetForSequenceClassification,m.MPNetForTokenClassification,m.MPNetModel,m.MPNetPreTrainedModel,m.MPNetTokenizer,m.MT5ForConditionalGeneration,m.MT5Model,m.MT5PreTrainedModel,m.MarianMTModel,m.MarianModel,m.MarianPreTrainedModel,m.MarianTokenizer,m.Mask2FormerImageProcessor,m.MaskFormerFeatureExtractor,m.MaskFormerForInstanceSegmentation,m.MaskFormerImageProcessor,m.MaskFormerModel,m.MaskFormerPreTrainedModel,m.MaskedLMOutput,m.MaxLengthCriteria,m.Metric3DForDepthEstimation,m.Metric3DPreTrainedModel,m.Metric3Dv2ForDepthEstimation,m.Metric3Dv2PreTrainedModel,m.MgpstrForSceneTextRecognition,m.MgpstrModelOutput,m.MgpstrPreTrainedModel,m.MgpstrProcessor,m.MgpstrTokenizer,m.MimiDecoderModel,m.MimiDecoderOutput,m.MimiEncoderModel,m.MimiEncoderOutput,m.MimiModel,m.MimiPreTrainedModel,m.MinLengthLogitsProcessor,m.MinNewTokensLengthLogitsProcessor,m.MistralForCausalLM,m.MistralModel,m.MistralPreTrainedModel,m.MobileBertForMaskedLM,m.MobileBertForQuestionAnswering,m.MobileBertForSequenceClassification,m.MobileBertModel,m.MobileBertPreTrainedModel,m.MobileBertTokenizer,m.MobileLLMForCausalLM,m.MobileLLMModel,m.MobileLLMPreTrainedModel,m.MobileNetV1FeatureExtractor,m.MobileNetV1ForImageClassification,m.MobileNetV1ForSemanticSegmentation,m.MobileNetV1ImageProcessor,m.MobileNetV1Model,m.MobileNetV1PreTrainedModel,m.MobileNetV2FeatureExtractor,m.MobileNetV2ForImageClassification,m.MobileNetV2ForSemanticSegmentation,m.MobileNetV2ImageProcessor,m.MobileNetV2Model,m.MobileNetV2PreTrainedModel,m.MobileNetV3FeatureExtractor,m.MobileNetV3ForImageClassification,m.MobileNetV3ForSemanticSegmentation,m.MobileNetV3ImageProcessor,m.MobileNetV3Model,m.MobileNetV3PreTrainedModel,m.MobileNetV4FeatureExtractor,m.MobileNetV4ForImageClassification,m.MobileNetV4ForSemanticSegmentation,m.MobileNetV4ImageProcessor,m.MobileNetV4Model,m.MobileNetV4PreTrainedModel,m.MobileViTFeatureExtractor,m.MobileViTForImageClassification,m.MobileViTImageProcessor,m.MobileViTModel,m.MobileViTPreTrainedModel,m.MobileViTV2ForImageClassification,m.MobileViTV2Model,m.MobileViTV2PreTrainedModel,m.ModelOutput,m.ModernBertDecoderForCausalLM,m.ModernBertDecoderModel,m.ModernBertDecoderPreTrainedModel,m.ModernBertForMaskedLM,m.ModernBertForSequenceClassification,m.ModernBertForTokenClassification,m.ModernBertModel,m.ModernBertPreTrainedModel,m.Moondream1ForConditionalGeneration,m.MoonshineFeatureExtractor,m.MoonshineForConditionalGeneration,m.MoonshineModel,m.MoonshinePreTrainedModel,m.MoonshineProcessor,m.MptForCausalLM,m.MptModel,m.MptPreTrainedModel,m.MultiModalityCausalLM,m.MultiModalityPreTrainedModel,m.MusicgenForCausalLM,m.MusicgenForConditionalGeneration,m.MusicgenModel,m.MusicgenPreTrainedModel,m.NanoChatForCausalLM,m.NanoChatModel,m.NanoChatPreTrainedModel,m.NeoBertForMaskedLM,m.NeoBertForQuestionAnswering,m.NeoBertForSequenceClassification,m.NeoBertForTokenClassification,m.NeoBertModel,m.NeoBertPreTrainedModel,m.NllbTokenizer,m.NoBadWordsLogitsProcessor,m.NoRepeatNGramLogitsProcessor,m.NomicBertModel,m.NomicBertPreTrainedModel,m.NougatImageProcessor,m.NougatTokenizer,m.OPTForCausalLM,m.OPTModel,m.OPTPreTrainedModel,m.ObjectDetectionPipeline,m.Olmo2ForCausalLM,m.Olmo2Model,m.Olmo2PreTrainedModel,m.OlmoForCausalLM,m.OlmoModel,m.OlmoPreTrainedModel,m.OpenELMForCausalLM,m.OpenELMModel,m.OpenELMPreTrainedModel,m.OwlViTFeatureExtractor,m.OwlViTForObjectDetection,m.OwlViTImageProcessor,m.OwlViTModel,m.OwlViTPreTrainedModel,m.OwlViTProcessor,m.Owlv2ForObjectDetection,m.Owlv2ImageProcessor,m.Owlv2Model,m.Owlv2PreTrainedModel,m.PaliGemmaForConditionalGeneration,m.PaliGemmaPreTrainedModel,m.PaliGemmaProcessor,m.ParakeetFeatureExtractor,m.ParakeetForCTC,m.ParakeetPreTrainedModel,m.PatchTSMixerForPrediction,m.PatchTSMixerModel,m.PatchTSMixerPreTrainedModel,m.PatchTSTForPrediction,m.PatchTSTModel,m.PatchTSTPreTrainedModel,m.Phi3ForCausalLM,m.Phi3Model,m.Phi3PreTrainedModel,m.Phi3VForCausalLM,m.Phi3VImageProcessor,m.Phi3VPreTrainedModel,m.Phi3VProcessor,m.PhiForCausalLM,m.PhiModel,m.PhiPreTrainedModel,m.Pipeline,m.PreTrainedModel,m.PreTrainedTokenizer,m.PretrainedConfig,m.PretrainedMixin,m.Processor,m.PvtForImageClassification,m.PvtImageProcessor,m.PvtModel,m.PvtPreTrainedModel,m.PyAnnoteFeatureExtractor,m.PyAnnoteForAudioFrameClassification,m.PyAnnoteModel,m.PyAnnotePreTrainedModel,m.PyAnnoteProcessor,m.QuestionAnsweringModelOutput,m.QuestionAnsweringPipeline,m.Qwen2ForCausalLM,m.Qwen2Model,m.Qwen2PreTrainedModel,m.Qwen2Tokenizer,m.Qwen2VLForConditionalGeneration,m.Qwen2VLImageProcessor,m.Qwen2VLPreTrainedModel,m.Qwen2VLProcessor,m.Qwen3ForCausalLM,m.Qwen3Model,m.Qwen3PreTrainedModel,m.RFDetrForObjectDetection,m.RFDetrModel,m.RFDetrObjectDetectionOutput,m.RFDetrPreTrainedModel,m.RTDetrForObjectDetection,m.RTDetrImageProcessor,m.RTDetrModel,m.RTDetrObjectDetectionOutput,m.RTDetrPreTrainedModel,m.RTDetrV2ForObjectDetection,m.RTDetrV2Model,m.RTDetrV2ObjectDetectionOutput,m.RTDetrV2PreTrainedModel,m.RawAudio;var KT=m.RawImage;m.RawVideo,m.RawVideoFrame,m.RepetitionPenaltyLogitsProcessor,m.ResNetForImageClassification,m.ResNetModel,m.ResNetPreTrainedModel,m.RoFormerForMaskedLM,m.RoFormerForQuestionAnswering,m.RoFormerForSequenceClassification,m.RoFormerForTokenClassification,m.RoFormerModel,m.RoFormerPreTrainedModel,m.RoFormerTokenizer,m.RobertaForMaskedLM,m.RobertaForQuestionAnswering,m.RobertaForSequenceClassification,m.RobertaForTokenClassification,m.RobertaModel,m.RobertaPreTrainedModel,m.RobertaTokenizer,m.SamImageProcessor,m.SamImageSegmentationOutput,m.SamModel,m.SamPreTrainedModel,m.SamProcessor,m.SapiensForDepthEstimation,m.SapiensForNormalEstimation,m.SapiensForSemanticSegmentation,m.SapiensPreTrainedModel,m.SeamlessM4TFeatureExtractor,m.SegformerFeatureExtractor,m.SegformerForImageClassification,m.SegformerForSemanticSegmentation,m.SegformerImageProcessor,m.SegformerModel,m.SegformerPreTrainedModel,m.Seq2SeqLMOutput,m.SequenceClassifierOutput,m.SiglipImageProcessor,m.SiglipModel,m.SiglipPreTrainedModel,m.SiglipTextModel,m.SiglipTokenizer,m.SiglipVisionModel,m.SmolLM3ForCausalLM,m.SmolLM3Model,m.SmolLM3PreTrainedModel,m.SmolVLMForConditionalGeneration,m.SmolVLMImageProcessor,m.SmolVLMProcessor,m.SnacDecoderModel,m.SnacEncoderModel,m.SnacFeatureExtractor,m.SnacModel,m.SnacPreTrainedModel,m.SpeechT5FeatureExtractor,m.SpeechT5ForSpeechToText,m.SpeechT5ForTextToSpeech,m.SpeechT5HifiGan,m.SpeechT5Model,m.SpeechT5PreTrainedModel,m.SpeechT5Processor,m.SpeechT5Tokenizer,m.SqueezeBertForMaskedLM,m.SqueezeBertForQuestionAnswering,m.SqueezeBertForSequenceClassification,m.SqueezeBertModel,m.SqueezeBertPreTrainedModel,m.SqueezeBertTokenizer,m.StableLmForCausalLM,m.StableLmModel,m.StableLmPreTrainedModel,m.Starcoder2ForCausalLM,m.Starcoder2Model,m.Starcoder2PreTrainedModel,m.StoppingCriteria,m.StoppingCriteriaList,m.StyleTextToSpeech2Model,m.StyleTextToSpeech2PreTrainedModel,m.SummarizationPipeline,m.SuppressTokensAtBeginLogitsProcessor,m.Swin2SRForImageSuperResolution,m.Swin2SRImageProcessor,m.Swin2SRModel,m.Swin2SRPreTrainedModel,m.SwinForImageClassification,m.SwinForSemanticSegmentation,m.SwinModel,m.SwinPreTrainedModel,m.T5ForConditionalGeneration,m.T5Model,m.T5PreTrainedModel,m.T5Tokenizer,m.TableTransformerForObjectDetection,m.TableTransformerModel,m.TableTransformerObjectDetectionOutput,m.TableTransformerPreTrainedModel,m.TemperatureLogitsWarper,m.Tensor,m.Text2TextGenerationPipeline,m.TextClassificationPipeline,m.TextGenerationPipeline,m.TextStreamer,m.TextToAudioPipeline,m.TokenClassificationPipeline,m.TokenClassifierOutput,m.TokenizerModel,m.TopKLogitsWarper,m.TopPLogitsWarper,m.TrOCRForCausalLM,m.TrOCRPreTrainedModel,m.TranslationPipeline,m.UltravoxModel,m.UltravoxPreTrainedModel,m.UltravoxProcessor,m.UniSpeechForCTC,m.UniSpeechForSequenceClassification,m.UniSpeechModel,m.UniSpeechPreTrainedModel,m.UniSpeechSatForAudioFrameClassification,m.UniSpeechSatForCTC,m.UniSpeechSatForSequenceClassification,m.UniSpeechSatModel,m.UniSpeechSatPreTrainedModel,m.VLChatProcessor,m.VLMImageProcessor,m.VaultGemmaForCausalLM,m.VaultGemmaModel,m.VaultGemmaPreTrainedModel,m.ViTFeatureExtractor,m.ViTForImageClassification,m.ViTImageProcessor,m.ViTMAEModel,m.ViTMAEPreTrainedModel,m.ViTMSNForImageClassification,m.ViTMSNModel,m.ViTMSNPreTrainedModel,m.ViTModel,m.ViTPreTrainedModel,m.VisionEncoderDecoderModel,m.VitMatteForImageMatting,m.VitMatteImageProcessor,m.VitMattePreTrainedModel,m.VitPoseForPoseEstimation,m.VitPoseImageProcessor,m.VitPosePreTrainedModel,m.VitsModel,m.VitsModelOutput,m.VitsPreTrainedModel,m.VitsTokenizer,m.VoxtralForConditionalGeneration,m.VoxtralProcessor,m.Wav2Vec2BertForCTC,m.Wav2Vec2BertForSequenceClassification,m.Wav2Vec2BertModel,m.Wav2Vec2BertPreTrainedModel,m.Wav2Vec2CTCTokenizer,m.Wav2Vec2FeatureExtractor,m.Wav2Vec2ForAudioFrameClassification,m.Wav2Vec2ForCTC,m.Wav2Vec2ForSequenceClassification,m.Wav2Vec2Model,m.Wav2Vec2PreTrainedModel,m.Wav2Vec2Processor,m.Wav2Vec2ProcessorWithLM,m.WavLMForAudioFrameClassification,m.WavLMForCTC,m.WavLMForSequenceClassification,m.WavLMForXVector,m.WavLMModel,m.WavLMPreTrainedModel,m.WeSpeakerFeatureExtractor,m.WeSpeakerResNetModel,m.WeSpeakerResNetPreTrainedModel,m.WhisperFeatureExtractor,m.WhisperForConditionalGeneration,m.WhisperModel,m.WhisperPreTrainedModel,m.WhisperProcessor,m.WhisperTextStreamer,m.WhisperTimeStampLogitsProcessor,m.WhisperTokenizer,m.XLMForQuestionAnswering,m.XLMForSequenceClassification,m.XLMForTokenClassification,m.XLMModel,m.XLMPreTrainedModel,m.XLMRobertaForMaskedLM,m.XLMRobertaForQuestionAnswering,m.XLMRobertaForSequenceClassification,m.XLMRobertaForTokenClassification,m.XLMRobertaModel,m.XLMRobertaPreTrainedModel,m.XLMRobertaTokenizer,m.XLMTokenizer,m.XLMWithLMHeadModel,m.XVectorOutput,m.YolosFeatureExtractor,m.YolosForObjectDetection,m.YolosImageProcessor,m.YolosModel,m.YolosObjectDetectionOutput,m.YolosPreTrainedModel,m.ZeroShotAudioClassificationPipeline,m.ZeroShotClassificationPipeline,m.ZeroShotImageClassificationPipeline,m.ZeroShotObjectDetectionPipeline,m.bankers_round,m.cat,m.cos_sim,m.dot,m.dynamic_time_warping,m.env,m.full,m.full_like,m.getCacheShapes,m.hamming,m.hanning,m.interpolate,m.interpolate_4d,m.interpolate_data,m.is_chinese_char,m.layer_norm,m.load_image,m.load_video,m.log_softmax,m.magnitude,m.matmul,m.max,m.mean,m.mean_pooling,m.medianFilter,m.mel_filter_bank,m.min,m.ones,m.ones_like,m.permute,m.permute_data,m.pipeline,m.quantize_embeddings,m.rand,m.read_audio,m.rfft,m.round,m.slice,m.softmax,m.spectrogram,m.stack,m.std_mean,m.topk,m.window_function,m.zeros,m.zeros_like;function qT(e){return e&&e.__esModule&&Object.prototype.hasOwnProperty.call(e,"default")?e.default:e}var bu={},yu,ww;function Ea(){if(ww)return yu;ww=1;const e="\\\\/",r=`[^${e}]`,t="\\.",s="\\+",o="\\?",n="\\/",a="(?=.)",i="[^/]",l=`(?:${n}|$)`,c=`(?:^|${n})`,p=`${t}{1,2}${l}`,d=`(?!${t})`,u=`(?!${c}${p})`,_=`(?!${t}{0,1}${l})`,f=`(?!${p})`,b=`[^.${n}]`,A=`${i}*?`,y={DOT_LITERAL:t,PLUS_LITERAL:s,QMARK_LITERAL:o,SLASH_LITERAL:n,ONE_CHAR:a,QMARK:i,END_ANCHOR:l,DOTS_SLASH:p,NO_DOT:d,NO_DOTS:u,NO_DOT_SLASH:_,NO_DOTS_SLASH:f,QMARK_NO_DOT:b,STAR:A,START_ANCHOR:c,SEP:"/"},C={...y,SLASH_LITERAL:`[${e}]`,QMARK:r,STAR:`${r}*?`,DOTS_SLASH:`${t}{1,2}(?:[${e}]|$)`,NO_DOT:`(?!${t})`,NO_DOTS:`(?!(?:^|[${e}])${t}{1,2}(?:[${e}]|$))`,NO_DOT_SLASH:`(?!${t}{0,1}(?:[${e}]|$))`,NO_DOTS_SLASH:`(?!${t}{1,2}(?:[${e}]|$))`,QMARK_NO_DOT:`[^.${e}]`,START_ANCHOR:`(?:^|[${e}])`,END_ANCHOR:`(?:[${e}]|$)`,SEP:"\\"},x={alnum:"a-zA-Z0-9",alpha:"a-zA-Z",ascii:"\\x00-\\x7F",blank:" \\t",cntrl:"\\x00-\\x1F\\x7F",digit:"0-9",graph:"\\x21-\\x7E",lower:"a-z",print:"\\x20-\\x7E ",punct:"\\-!\"#$%&'()\\*+,./:;<=>?@[\\]^_`{|}~",space:" \\t\\r\\n\\v\\f",upper:"A-Z",word:"A-Za-z0-9_",xdigit:"A-Fa-f0-9"};return yu={MAX_LENGTH:1024*64,POSIX_REGEX_SOURCE:x,REGEX_BACKSLASH:/\\(?![*+?^${}(|)[\]])/g,REGEX_NON_SPECIAL_CHARS:/^[^@![\].,$*+?^{}()|\\/]+/,REGEX_SPECIAL_CHARS:/[-*+?.^${}(|)[\]]/,REGEX_SPECIAL_CHARS_BACKREF:/(\\?)((\W)(\3*))/g,REGEX_SPECIAL_CHARS_GLOBAL:/([-*+?.^${}(|)[\]])/g,REGEX_REMOVE_BACKSLASH:/(?:\[.*?[^\\]\]|\\(?=.))/g,REPLACEMENTS:{__proto__:null,"***":"*","**/**":"**","**/**/**":"**"},CHAR_0:48,CHAR_9:57,CHAR_UPPERCASE_A:65,CHAR_LOWERCASE_A:97,CHAR_UPPERCASE_Z:90,CHAR_LOWERCASE_Z:122,CHAR_LEFT_PARENTHESES:40,CHAR_RIGHT_PARENTHESES:41,CHAR_ASTERISK:42,CHAR_AMPERSAND:38,CHAR_AT:64,CHAR_BACKWARD_SLASH:92,CHAR_CARRIAGE_RETURN:13,CHAR_CIRCUMFLEX_ACCENT:94,CHAR_COLON:58,CHAR_COMMA:44,CHAR_DOT:46,CHAR_DOUBLE_QUOTE:34,CHAR_EQUAL:61,CHAR_EXCLAMATION_MARK:33,CHAR_FORM_FEED:12,CHAR_FORWARD_SLASH:47,CHAR_GRAVE_ACCENT:96,CHAR_HASH:35,CHAR_HYPHEN_MINUS:45,CHAR_LEFT_ANGLE_BRACKET:60,CHAR_LEFT_CURLY_BRACE:123,CHAR_LEFT_SQUARE_BRACKET:91,CHAR_LINE_FEED:10,CHAR_NO_BREAK_SPACE:160,CHAR_PERCENT:37,CHAR_PLUS:43,CHAR_QUESTION_MARK:63,CHAR_RIGHT_ANGLE_BRACKET:62,CHAR_RIGHT_CURLY_BRACE:125,CHAR_RIGHT_SQUARE_BRACKET:93,CHAR_SEMICOLON:59,CHAR_SINGLE_QUOTE:39,CHAR_SPACE:32,CHAR_TAB:9,CHAR_UNDERSCORE:95,CHAR_VERTICAL_LINE:124,CHAR_ZERO_WIDTH_NOBREAK_SPACE:65279,extglobChars(M){return{"!":{type:"negate",open:"(?:(?!(?:",close:`))${M.STAR})`},"?":{type:"qmark",open:"(?:",close:")?"},"+":{type:"plus",open:"(?:",close:")+"},"*":{type:"star",open:"(?:",close:")*"},"@":{type:"at",open:"(?:",close:")"}}},globChars(M){return M===!0?C:y}},yu}var Mw;function Pa(){return Mw||(Mw=1,function(e){const{REGEX_BACKSLASH:r,REGEX_REMOVE_BACKSLASH:t,REGEX_SPECIAL_CHARS:s,REGEX_SPECIAL_CHARS_GLOBAL:o}=Ea();e.isObject=n=>n!==null&&typeof n=="object"&&!Array.isArray(n),e.hasRegexChars=n=>s.test(n),e.isRegexChar=n=>n.length===1&&e.hasRegexChars(n),e.escapeRegex=n=>n.replace(o,"\\$1"),e.toPosixSlashes=n=>n.replace(r,"/"),e.isWindows=()=>{if(typeof navigator<"u"&&navigator.platform){const n=navigator.platform.toLowerCase();return n==="win32"||n==="windows"}return typeof process<"u"&&process.platform?process.platform==="win32":!1},e.removeBackslashes=n=>n.replace(t,a=>a==="\\"?"":a),e.escapeLast=(n,a,i)=>{const l=n.lastIndexOf(a,i);return l===-1?n:n[l-1]==="\\"?e.escapeLast(n,a,l-1):`${n.slice(0,l)}\\${n.slice(l)}`},e.removePrefix=(n,a={})=>{let i=n;return i.startsWith("./")&&(i=i.slice(2),a.prefix="./"),i},e.wrapOutput=(n,a={},i={})=>{const l=i.contains?"":"^",c=i.contains?"":"$";let p=`${l}(?:${n})${c}`;return a.negated===!0&&(p=`(?:^(?!${p}).*$)`),p},e.basename=(n,{windows:a}={})=>{const i=n.split(a?/[\\/]/:"/"),l=i[i.length-1];return l===""?i[i.length-2]:l}}(bu)),bu}var vu,bw;function QT(){if(bw)return vu;bw=1;const e=Pa(),{CHAR_ASTERISK:r,CHAR_AT:t,CHAR_BACKWARD_SLASH:s,CHAR_COMMA:o,CHAR_DOT:n,CHAR_EXCLAMATION_MARK:a,CHAR_FORWARD_SLASH:i,CHAR_LEFT_CURLY_BRACE:l,CHAR_LEFT_PARENTHESES:c,CHAR_LEFT_SQUARE_BRACKET:p,CHAR_PLUS:d,CHAR_QUESTION_MARK:u,CHAR_RIGHT_CURLY_BRACE:_,CHAR_RIGHT_PARENTHESES:f,CHAR_RIGHT_SQUARE_BRACKET:b}=Ea(),A=C=>C===i||C===s,g=C=>{C.isPrefix!==!0&&(C.depth=C.isGlobstar?1/0:1)};return vu=(C,x)=>{const M=x||{},T=C.length-1,v=M.parts===!0||M.scanToEnd===!0,P=[],F=[],D=[];let K=C,U=-1,j=0,ne=0,q=!1,te=!1,Z=!1,ae=!1,he=!1,Q=!1,B=!1,O=!1,W=!1,N=!1,J=0,ie,me,Ae={value:"",depth:0,isGlob:!1};const Ve=()=>U>=T,$e=()=>K.charCodeAt(U+1),X=()=>(ie=me,K.charCodeAt(++U));for(;U<T;){me=X();let Ce;if(me===s){B=Ae.backslashes=!0,me=X(),me===l&&(Q=!0);continue}if(Q===!0||me===l){for(J++;Ve()!==!0&&(me=X());){if(me===s){B=Ae.backslashes=!0,X();continue}if(me===l){J++;continue}if(Q!==!0&&me===n&&(me=X())===n){if(q=Ae.isBrace=!0,Z=Ae.isGlob=!0,N=!0,v===!0)continue;break}if(Q!==!0&&me===o){if(q=Ae.isBrace=!0,Z=Ae.isGlob=!0,N=!0,v===!0)continue;break}if(me===_&&(J--,J===0)){Q=!1,q=Ae.isBrace=!0,N=!0;break}}if(v===!0)continue;break}if(me===i){if(P.push(U),F.push(Ae),Ae={value:"",depth:0,isGlob:!1},N===!0)continue;if(ie===n&&U===j+1){j+=2;continue}ne=U+1;continue}if(M.noext!==!0&&(me===d||me===t||me===r||me===u||me===a)===!0&&$e()===c){if(Z=Ae.isGlob=!0,ae=Ae.isExtglob=!0,N=!0,me===a&&U===j&&(W=!0),v===!0){for(;Ve()!==!0&&(me=X());){if(me===s){B=Ae.backslashes=!0,me=X();continue}if(me===f){Z=Ae.isGlob=!0,N=!0;break}}continue}break}if(me===r){if(ie===r&&(he=Ae.isGlobstar=!0),Z=Ae.isGlob=!0,N=!0,v===!0)continue;break}if(me===u){if(Z=Ae.isGlob=!0,N=!0,v===!0)continue;break}if(me===p){for(;Ve()!==!0&&(Ce=X());){if(Ce===s){B=Ae.backslashes=!0,X();continue}if(Ce===b){te=Ae.isBracket=!0,Z=Ae.isGlob=!0,N=!0;break}}if(v===!0)continue;break}if(M.nonegate!==!0&&me===a&&U===j){O=Ae.negated=!0,j++;continue}if(M.noparen!==!0&&me===c){if(Z=Ae.isGlob=!0,v===!0){for(;Ve()!==!0&&(me=X());){if(me===c){B=Ae.backslashes=!0,me=X();continue}if(me===f){N=!0;break}}continue}break}if(Z===!0){if(N=!0,v===!0)continue;break}}M.noext===!0&&(ae=!1,Z=!1);let z=K,_e="",Ee="";j>0&&(_e=K.slice(0,j),K=K.slice(j),ne-=j),z&&Z===!0&&ne>0?(z=K.slice(0,ne),Ee=K.slice(ne)):Z===!0?(z="",Ee=K):z=K,z&&z!==""&&z!=="/"&&z!==K&&A(z.charCodeAt(z.length-1))&&(z=z.slice(0,-1)),M.unescape===!0&&(Ee&&(Ee=e.removeBackslashes(Ee)),z&&B===!0&&(z=e.removeBackslashes(z)));const Me={prefix:_e,input:C,start:j,base:z,glob:Ee,isBrace:q,isBracket:te,isGlob:Z,isExtglob:ae,isGlobstar:he,negated:O,negatedExtglob:W};if(M.tokens===!0&&(Me.maxDepth=0,A(me)||F.push(Ae),Me.tokens=F),M.parts===!0||M.tokens===!0){let Ce;for(let ye=0;ye<P.length;ye++){const de=Ce?Ce+1:j,we=P[ye],ce=C.slice(de,we);M.tokens&&(ye===0&&j!==0?(F[ye].isPrefix=!0,F[ye].value=_e):F[ye].value=ce,g(F[ye]),Me.maxDepth+=F[ye].depth),(ye!==0||ce!=="")&&D.push(ce),Ce=we}if(Ce&&Ce+1<C.length){const ye=C.slice(Ce+1);D.push(ye),M.tokens&&(F[F.length-1].value=ye,g(F[F.length-1]),Me.maxDepth+=F[F.length-1].depth)}Me.slashes=P,Me.parts=D}return Me},vu}var xu,yw;function XT(){if(yw)return xu;yw=1;const e=Ea(),r=Pa(),{MAX_LENGTH:t,POSIX_REGEX_SOURCE:s,REGEX_NON_SPECIAL_CHARS:o,REGEX_SPECIAL_CHARS_BACKREF:n,REPLACEMENTS:a}=e,i=(p,d)=>{if(typeof d.expandRange=="function")return d.expandRange(...p,d);p.sort();const u=`[${p.join("-")}]`;try{new RegExp(u)}catch{return p.map(f=>r.escapeRegex(f)).join("..")}return u},l=(p,d)=>`Missing ${p}: "${d}" - use "\\\\${d}" to match literal characters`,c=(p,d)=>{if(typeof p!="string")throw new TypeError("Expected a string");p=a[p]||p;const u={...d},_=typeof u.maxLength=="number"?Math.min(t,u.maxLength):t;let f=p.length;if(f>_)throw new SyntaxError(`Input length: ${f}, exceeds maximum allowed length: ${_}`);const b={type:"bos",value:"",output:u.prepend||""},A=[b],g=u.capture?"":"?:",y=e.globChars(u.windows),C=e.extglobChars(y),{DOT_LITERAL:x,PLUS_LITERAL:M,SLASH_LITERAL:T,ONE_CHAR:v,DOTS_SLASH:P,NO_DOT:F,NO_DOT_SLASH:D,NO_DOTS_SLASH:K,QMARK:U,QMARK_NO_DOT:j,STAR:ne,START_ANCHOR:q}=y,te=de=>`(${g}(?:(?!${q}${de.dot?P:x}).)*?)`,Z=u.dot?"":F,ae=u.dot?U:j;let he=u.bash===!0?te(u):ne;u.capture&&(he=`(${he})`),typeof u.noext=="boolean"&&(u.noextglob=u.noext);const Q={input:p,index:-1,start:0,dot:u.dot===!0,consumed:"",output:"",prefix:"",backtrack:!1,negated:!1,brackets:0,braces:0,parens:0,quotes:0,globstar:!1,tokens:A};p=r.removePrefix(p,Q),f=p.length;const B=[],O=[],W=[];let N=b,J;const ie=()=>Q.index===f-1,me=Q.peek=(de=1)=>p[Q.index+de],Ae=Q.advance=()=>p[++Q.index]||"",Ve=()=>p.slice(Q.index+1),$e=(de="",we=0)=>{Q.consumed+=de,Q.index+=we},X=de=>{Q.output+=de.output!=null?de.output:de.value,$e(de.value)},z=()=>{let de=1;for(;me()==="!"&&(me(2)!=="("||me(3)==="?");)Ae(),Q.start++,de++;return de%2===0?!1:(Q.negated=!0,Q.start++,!0)},_e=de=>{Q[de]++,W.push(de)},Ee=de=>{Q[de]--,W.pop()},Me=de=>{if(N.type==="globstar"){const we=Q.braces>0&&(de.type==="comma"||de.type==="brace"),ce=de.extglob===!0||B.length&&(de.type==="pipe"||de.type==="paren");de.type!=="slash"&&de.type!=="paren"&&!we&&!ce&&(Q.output=Q.output.slice(0,-N.output.length),N.type="star",N.value="*",N.output=he,Q.output+=N.output)}if(B.length&&de.type!=="paren"&&(B[B.length-1].inner+=de.value),(de.value||de.output)&&X(de),N&&N.type==="text"&&de.type==="text"){N.output=(N.output||N.value)+de.value,N.value+=de.value;return}de.prev=N,A.push(de),N=de},Ce=(de,we)=>{const ce={...C[we],conditions:1,inner:""};ce.prev=N,ce.parens=Q.parens,ce.output=Q.output;const ke=(u.capture?"(":"")+ce.open;_e("parens"),Me({type:de,value:we,output:Q.output?"":v}),Me({type:"paren",extglob:!0,value:Ae(),output:ke}),B.push(ce)},ye=de=>{let we=de.close+(u.capture?")":""),ce;if(de.type==="negate"){let ke=he;if(de.inner&&de.inner.length>1&&de.inner.includes("/")&&(ke=te(u)),(ke!==he||ie()||/^\)+$/.test(Ve()))&&(we=de.close=`)$))${ke}`),de.inner.includes("*")&&(ce=Ve())&&/^\.[^\\/.]+$/.test(ce)){const Le=c(ce,{...d,fastpaths:!1}).output;we=de.close=`)${Le})${ke})`}de.prev.type==="bos"&&(Q.negatedExtglob=!0)}Me({type:"paren",extglob:!0,value:J,output:we}),Ee("parens")};if(u.fastpaths!==!1&&!/(^[*!]|[/()[\]{}"])/.test(p)){let de=!1,we=p.replace(n,(ce,ke,Le,Te,We,qe)=>Te==="\\"?(de=!0,ce):Te==="?"?ke?ke+Te+(We?U.repeat(We.length):""):qe===0?ae+(We?U.repeat(We.length):""):U.repeat(Le.length):Te==="."?x.repeat(Le.length):Te==="*"?ke?ke+Te+(We?he:""):he:ke?ce:`\\${ce}`);return de===!0&&(u.unescape===!0?we=we.replace(/\\/g,""):we=we.replace(/\\+/g,ce=>ce.length%2===0?"\\\\":ce?"\\":"")),we===p&&u.contains===!0?(Q.output=p,Q):(Q.output=r.wrapOutput(we,Q,d),Q)}for(;!ie();){if(J=Ae(),J==="\0")continue;if(J==="\\"){const ce=me();if(ce==="/"&&u.bash!==!0||ce==="."||ce===";")continue;if(!ce){J+="\\",Me({type:"text",value:J});continue}const ke=/^\\+/.exec(Ve());let Le=0;if(ke&&ke[0].length>2&&(Le=ke[0].length,Q.index+=Le,Le%2!==0&&(J+="\\")),u.unescape===!0?J=Ae():J+=Ae(),Q.brackets===0){Me({type:"text",value:J});continue}}if(Q.brackets>0&&(J!=="]"||N.value==="["||N.value==="[^")){if(u.posix!==!1&&J===":"){const ce=N.value.slice(1);if(ce.includes("[")&&(N.posix=!0,ce.includes(":"))){const ke=N.value.lastIndexOf("["),Le=N.value.slice(0,ke),Te=N.value.slice(ke+2),We=s[Te];if(We){N.value=Le+We,Q.backtrack=!0,Ae(),!b.output&&A.indexOf(N)===1&&(b.output=v);continue}}}(J==="["&&me()!==":"||J==="-"&&me()==="]")&&(J=`\\${J}`),J==="]"&&(N.value==="["||N.value==="[^")&&(J=`\\${J}`),u.posix===!0&&J==="!"&&N.value==="["&&(J="^"),N.value+=J,X({value:J});continue}if(Q.quotes===1&&J!=='"'){J=r.escapeRegex(J),N.value+=J,X({value:J});continue}if(J==='"'){Q.quotes=Q.quotes===1?0:1,u.keepQuotes===!0&&Me({type:"text",value:J});continue}if(J==="("){_e("parens"),Me({type:"paren",value:J});continue}if(J===")"){if(Q.parens===0&&u.strictBrackets===!0)throw new SyntaxError(l("opening","("));const ce=B[B.length-1];if(ce&&Q.parens===ce.parens+1){ye(B.pop());continue}Me({type:"paren",value:J,output:Q.parens?")":"\\)"}),Ee("parens");continue}if(J==="["){if(u.nobracket===!0||!Ve().includes("]")){if(u.nobracket!==!0&&u.strictBrackets===!0)throw new SyntaxError(l("closing","]"));J=`\\${J}`}else _e("brackets");Me({type:"bracket",value:J});continue}if(J==="]"){if(u.nobracket===!0||N&&N.type==="bracket"&&N.value.length===1){Me({type:"text",value:J,output:`\\${J}`});continue}if(Q.brackets===0){if(u.strictBrackets===!0)throw new SyntaxError(l("opening","["));Me({type:"text",value:J,output:`\\${J}`});continue}Ee("brackets");const ce=N.value.slice(1);if(N.posix!==!0&&ce[0]==="^"&&!ce.includes("/")&&(J=`/${J}`),N.value+=J,X({value:J}),u.literalBrackets===!1||r.hasRegexChars(ce))continue;const ke=r.escapeRegex(N.value);if(Q.output=Q.output.slice(0,-N.value.length),u.literalBrackets===!0){Q.output+=ke,N.value=ke;continue}N.value=`(${g}${ke}|${N.value})`,Q.output+=N.value;continue}if(J==="{"&&u.nobrace!==!0){_e("braces");const ce={type:"brace",value:J,output:"(",outputIndex:Q.output.length,tokensIndex:Q.tokens.length};O.push(ce),Me(ce);continue}if(J==="}"){const ce=O[O.length-1];if(u.nobrace===!0||!ce){Me({type:"text",value:J,output:J});continue}let ke=")";if(ce.dots===!0){const Le=A.slice(),Te=[];for(let We=Le.length-1;We>=0&&(A.pop(),Le[We].type!=="brace");We--)Le[We].type!=="dots"&&Te.unshift(Le[We].value);ke=i(Te,u),Q.backtrack=!0}if(ce.comma!==!0&&ce.dots!==!0){const Le=Q.output.slice(0,ce.outputIndex),Te=Q.tokens.slice(ce.tokensIndex);ce.value=ce.output="\\{",J=ke="\\}",Q.output=Le;for(const We of Te)Q.output+=We.output||We.value}Me({type:"brace",value:J,output:ke}),Ee("braces"),O.pop();continue}if(J==="|"){B.length>0&&B[B.length-1].conditions++,Me({type:"text",value:J});continue}if(J===","){let ce=J;const ke=O[O.length-1];ke&&W[W.length-1]==="braces"&&(ke.comma=!0,ce="|"),Me({type:"comma",value:J,output:ce});continue}if(J==="/"){if(N.type==="dot"&&Q.index===Q.start+1){Q.start=Q.index+1,Q.consumed="",Q.output="",A.pop(),N=b;continue}Me({type:"slash",value:J,output:T});continue}if(J==="."){if(Q.braces>0&&N.type==="dot"){N.value==="."&&(N.output=x);const ce=O[O.length-1];N.type="dots",N.output+=J,N.value+=J,ce.dots=!0;continue}if(Q.braces+Q.parens===0&&N.type!=="bos"&&N.type!=="slash"){Me({type:"text",value:J,output:x});continue}Me({type:"dot",value:J,output:x});continue}if(J==="?"){if(!(N&&N.value==="(")&&u.noextglob!==!0&&me()==="("&&me(2)!=="?"){Ce("qmark",J);continue}if(N&&N.type==="paren"){const ke=me();let Le=J;(N.value==="("&&!/[!=<:]/.test(ke)||ke==="<"&&!/<([!=]|\w+>)/.test(Ve()))&&(Le=`\\${J}`),Me({type:"text",value:J,output:Le});continue}if(u.dot!==!0&&(N.type==="slash"||N.type==="bos")){Me({type:"qmark",value:J,output:j});continue}Me({type:"qmark",value:J,output:U});continue}if(J==="!"){if(u.noextglob!==!0&&me()==="("&&(me(2)!=="?"||!/[!=<:]/.test(me(3)))){Ce("negate",J);continue}if(u.nonegate!==!0&&Q.index===0){z();continue}}if(J==="+"){if(u.noextglob!==!0&&me()==="("&&me(2)!=="?"){Ce("plus",J);continue}if(N&&N.value==="("||u.regex===!1){Me({type:"plus",value:J,output:M});continue}if(N&&(N.type==="bracket"||N.type==="paren"||N.type==="brace")||Q.parens>0){Me({type:"plus",value:J});continue}Me({type:"plus",value:M});continue}if(J==="@"){if(u.noextglob!==!0&&me()==="("&&me(2)!=="?"){Me({type:"at",extglob:!0,value:J,output:""});continue}Me({type:"text",value:J});continue}if(J!=="*"){(J==="$"||J==="^")&&(J=`\\${J}`);const ce=o.exec(Ve());ce&&(J+=ce[0],Q.index+=ce[0].length),Me({type:"text",value:J});continue}if(N&&(N.type==="globstar"||N.star===!0)){N.type="star",N.star=!0,N.value+=J,N.output=he,Q.backtrack=!0,Q.globstar=!0,$e(J);continue}let de=Ve();if(u.noextglob!==!0&&/^\([^?]/.test(de)){Ce("star",J);continue}if(N.type==="star"){if(u.noglobstar===!0){$e(J);continue}const ce=N.prev,ke=ce.prev,Le=ce.type==="slash"||ce.type==="bos",Te=ke&&(ke.type==="star"||ke.type==="globstar");if(u.bash===!0&&(!Le||de[0]&&de[0]!=="/")){Me({type:"star",value:J,output:""});continue}const We=Q.braces>0&&(ce.type==="comma"||ce.type==="brace"),qe=B.length&&(ce.type==="pipe"||ce.type==="paren");if(!Le&&ce.type!=="paren"&&!We&&!qe){Me({type:"star",value:J,output:""});continue}for(;de.slice(0,3)==="/**";){const st=p[Q.index+4];if(st&&st!=="/")break;de=de.slice(3),$e("/**",3)}if(ce.type==="bos"&&ie()){N.type="globstar",N.value+=J,N.output=te(u),Q.output=N.output,Q.globstar=!0,$e(J);continue}if(ce.type==="slash"&&ce.prev.type!=="bos"&&!Te&&ie()){Q.output=Q.output.slice(0,-(ce.output+N.output).length),ce.output=`(?:${ce.output}`,N.type="globstar",N.output=te(u)+(u.strictSlashes?")":"|$)"),N.value+=J,Q.globstar=!0,Q.output+=ce.output+N.output,$e(J);continue}if(ce.type==="slash"&&ce.prev.type!=="bos"&&de[0]==="/"){const st=de[1]!==void 0?"|$":"";Q.output=Q.output.slice(0,-(ce.output+N.output).length),ce.output=`(?:${ce.output}`,N.type="globstar",N.output=`${te(u)}${T}|${T}${st})`,N.value+=J,Q.output+=ce.output+N.output,Q.globstar=!0,$e(J+Ae()),Me({type:"slash",value:"/",output:""});continue}if(ce.type==="bos"&&de[0]==="/"){N.type="globstar",N.value+=J,N.output=`(?:^|${T}|${te(u)}${T})`,Q.output=N.output,Q.globstar=!0,$e(J+Ae()),Me({type:"slash",value:"/",output:""});continue}Q.output=Q.output.slice(0,-N.output.length),N.type="globstar",N.output=te(u),N.value+=J,Q.output+=N.output,Q.globstar=!0,$e(J);continue}const we={type:"star",value:J,output:he};if(u.bash===!0){we.output=".*?",(N.type==="bos"||N.type==="slash")&&(we.output=Z+we.output),Me(we);continue}if(N&&(N.type==="bracket"||N.type==="paren")&&u.regex===!0){we.output=J,Me(we);continue}(Q.index===Q.start||N.type==="slash"||N.type==="dot")&&(N.type==="dot"?(Q.output+=D,N.output+=D):u.dot===!0?(Q.output+=K,N.output+=K):(Q.output+=Z,N.output+=Z),me()!=="*"&&(Q.output+=v,N.output+=v)),Me(we)}for(;Q.brackets>0;){if(u.strictBrackets===!0)throw new SyntaxError(l("closing","]"));Q.output=r.escapeLast(Q.output,"["),Ee("brackets")}for(;Q.parens>0;){if(u.strictBrackets===!0)throw new SyntaxError(l("closing",")"));Q.output=r.escapeLast(Q.output,"("),Ee("parens")}for(;Q.braces>0;){if(u.strictBrackets===!0)throw new SyntaxError(l("closing","}"));Q.output=r.escapeLast(Q.output,"{"),Ee("braces")}if(u.strictSlashes!==!0&&(N.type==="star"||N.type==="bracket")&&Me({type:"maybe_slash",value:"",output:`${T}?`}),Q.backtrack===!0){Q.output="";for(const de of Q.tokens)Q.output+=de.output!=null?de.output:de.value,de.suffix&&(Q.output+=de.suffix)}return Q};return c.fastpaths=(p,d)=>{const u={...d},_=typeof u.maxLength=="number"?Math.min(t,u.maxLength):t,f=p.length;if(f>_)throw new SyntaxError(`Input length: ${f}, exceeds maximum allowed length: ${_}`);p=a[p]||p;const{DOT_LITERAL:b,SLASH_LITERAL:A,ONE_CHAR:g,DOTS_SLASH:y,NO_DOT:C,NO_DOTS:x,NO_DOTS_SLASH:M,STAR:T,START_ANCHOR:v}=e.globChars(u.windows),P=u.dot?x:C,F=u.dot?M:C,D=u.capture?"":"?:",K={negated:!1,prefix:""};let U=u.bash===!0?".*?":T;u.capture&&(U=`(${U})`);const j=Z=>Z.noglobstar===!0?U:`(${D}(?:(?!${v}${Z.dot?y:b}).)*?)`,ne=Z=>{switch(Z){case"*":return`${P}${g}${U}`;case".*":return`${b}${g}${U}`;case"*.*":return`${P}${U}${b}${g}${U}`;case"*/*":return`${P}${U}${A}${g}${F}${U}`;case"**":return P+j(u);case"**/*":return`(?:${P}${j(u)}${A})?${F}${g}${U}`;case"**/*.*":return`(?:${P}${j(u)}${A})?${F}${U}${b}${g}${U}`;case"**/.*":return`(?:${P}${j(u)}${A})?${b}${g}${U}`;default:{const ae=/^(.*?)\.(\w+)$/.exec(Z);if(!ae)return;const he=ne(ae[1]);return he?he+b+ae[2]:void 0}}},q=r.removePrefix(p,K);let te=ne(q);return te&&u.strictSlashes!==!0&&(te+=`${A}?`),te},xu=c,xu}var Tu,vw;function JT(){if(vw)return Tu;vw=1;const e=QT(),r=XT(),t=Pa(),s=Ea(),o=a=>a&&typeof a=="object"&&!Array.isArray(a),n=(a,i,l=!1)=>{if(Array.isArray(a)){const A=a.map(y=>n(y,i,l));return y=>{for(const C of A){const x=C(y);if(x)return x}return!1}}const c=o(a)&&a.tokens&&a.input;if(a===""||typeof a!="string"&&!c)throw new TypeError("Expected pattern to be a non-empty string");const p=i||{},d=p.windows,u=c?n.compileRe(a,i):n.makeRe(a,i,!1,!0),_=u.state;delete u.state;let f=()=>!1;if(p.ignore){const A={...i,ignore:null,onMatch:null,onResult:null};f=n(p.ignore,A,l)}const b=(A,g=!1)=>{const{isMatch:y,match:C,output:x}=n.test(A,u,i,{glob:a,posix:d}),M={glob:a,state:_,regex:u,posix:d,input:A,output:x,match:C,isMatch:y};return typeof p.onResult=="function"&&p.onResult(M),y===!1?(M.isMatch=!1,g?M:!1):f(A)?(typeof p.onIgnore=="function"&&p.onIgnore(M),M.isMatch=!1,g?M:!1):(typeof p.onMatch=="function"&&p.onMatch(M),g?M:!0)};return l&&(b.state=_),b};return n.test=(a,i,l,{glob:c,posix:p}={})=>{if(typeof a!="string")throw new TypeError("Expected input to be a string");if(a==="")return{isMatch:!1,output:""};const d=l||{},u=d.format||(p?t.toPosixSlashes:null);let _=a===c,f=_&&u?u(a):a;return _===!1&&(f=u?u(a):a,_=f===c),(_===!1||d.capture===!0)&&(d.matchBase===!0||d.basename===!0?_=n.matchBase(a,i,l,p):_=i.exec(f)),{isMatch:!!_,match:_,output:f}},n.matchBase=(a,i,l)=>(i instanceof RegExp?i:n.makeRe(i,l)).test(t.basename(a)),n.isMatch=(a,i,l)=>n(i,l)(a),n.parse=(a,i)=>Array.isArray(a)?a.map(l=>n.parse(l,i)):r(a,{...i,fastpaths:!1}),n.scan=(a,i)=>e(a,i),n.compileRe=(a,i,l=!1,c=!1)=>{if(l===!0)return a.output;const p=i||{},d=p.contains?"":"^",u=p.contains?"":"$";let _=`${d}(?:${a.output})${u}`;a&&a.negated===!0&&(_=`^(?!${_}).*$`);const f=n.toRegex(_,i);return c===!0&&(f.state=a),f},n.makeRe=(a,i={},l=!1,c=!1)=>{if(!a||typeof a!="string")throw new TypeError("Expected a non-empty string");let p={negated:!1,fastpaths:!0};return i.fastpaths!==!1&&(a[0]==="."||a[0]==="*")&&(p.output=r.fastpaths(a,i)),p.output||(p=r(a,i)),n.compileRe(p,i,l,c)},n.toRegex=(a,i)=>{try{const l=i||{};return new RegExp(a,l.flags||(l.nocase?"i":""))}catch(l){if(i&&i.debug===!0)throw l;return/$^/}},n.constants=s,Tu=n,Tu}var Eu,xw;function YT(){if(xw)return Eu;xw=1;const e=JT(),r=Pa();function t(s,o,n=!1){return o&&(o.windows===null||o.windows===void 0)&&(o={...o,windows:r.isWindows()}),e(s,o,n)}return Object.assign(t,e),Eu=t,Eu}var ZT=YT(),eE=qT(ZT);let tE=e=>crypto.getRandomValues(new Uint8Array(e)),rE=(e,r,t)=>{let s=(2<<Math.log2(e.length-1))-1,o=-~(1.6*s*r/e.length);return(n=r)=>{let a="";for(;;){let i=t(o),l=o|0;for(;l--;)if(a+=e[i[l]&s]||"",a.length>=n)return a}}},sE=(e,r=21)=>rE(e,r|0,tE);function So(){return sE("0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz",16)()}let Ws=function(e){return e.Event="event",e.MatchExpression="matchExpression",e}({}),Pu=function(e){return e.Inbound="inbound",e.Outbound="outbound",e}({});function Tw(e){return{...cs(e),_flowDirection:Pu.Inbound}}function nE(e){return{...cs(e),_flowDirection:Pu.Outbound}}function cs(e){return e||(e=So()),{id:e,type:Ws.Event}}function oE(...e){return{id:So(),type:Ws.MatchExpression,matcher:r=>e.every(t=>t.matcher?t.matcher(r):!1)}}function Ew(e,r){const t=So();let s=()=>!1;return typeof e=="string"?s=o=>eE(e)(o.id):typeof e=="object"?"ids"in e?s=o=>e.ids.includes(o.id):"eventa"in e?s=o=>e.eventa.some(n=>n.id===o.id):"types"in e&&(s=o=>typeof o.type>"u"?!1:e.types.includes(o.type)):e instanceof RegExp?s=o=>e.test(o.id):typeof e=="function"&&(s=e),{id:t,type:Ws.MatchExpression,matcher:s}}function aE(e={}){var l;const r=new Map,t=new Map,s=new Map,o=new Map,n=new Map,a=(l=e.adapter)==null?void 0:l.call(e,i).hooks;function i(c,p,d){var _,f,b,A,g,y;const u={...c,body:p};for(const C of r.get(c.id)||[])C(u,d),(_=a==null?void 0:a.onReceived)==null||_.call(a,c.id,u);for(const C of t.get(c.id)||[])C(u,d),(f=a==null?void 0:a.onReceived)==null||f.call(a,c.id,u),(b=t.get(c.id))==null||b.delete(C);for(const C of s.values())if(C.matcher){if(!C.matcher(u))continue;for(const x of o.get(C.id)||[])x(u,d),(A=a==null?void 0:a.onReceived)==null||A.call(a,C.id,u);for(const x of n.get(C.id)||[])x(u,d),(g=a==null?void 0:a.onReceived)==null||g.call(a,C.id,u),(y=n.get(C.id))==null||y.delete(x)}a==null||a.onSent(c.id,u,d)}return{get listeners(){return r},get onceListeners(){return t},emit:i,on(c,p){var d,u;if(c.type===Ws.Event){const _=c;return r.has(_.id)||r.set(_.id,new Set),(d=r.get(_.id))==null||d.add(p),()=>{var f;return(f=r.get(_.id))==null?void 0:f.delete(p)}}if(c.type===Ws.MatchExpression){const _=c;return s.has(_.id)||s.set(_.id,_),o.has(_.id)||o.set(_.id,new Set),(u=o.get(_.id))==null||u.add(p),()=>{var f;return(f=o.get(_.id))==null?void 0:f.delete(p)}}return()=>{}},once(c,p){var d,u;if(c.type===Ws.Event){const _=c;return t.has(_.id)||t.set(_.id,new Set),(d=t.get(_.id))==null||d.add(p),()=>{var f;return(f=t.get(_.id))==null?void 0:f.delete(p)}}if(c.type===Ws.MatchExpression){const _=c;return s.has(_.id)||s.set(_.id,_),o.has(_.id)||o.set(_.id,new Set),(u=n.get(_.id))==null||u.add(p),()=>{var f;return(f=n.get(_.id))==null?void 0:f.delete(p)}}return()=>{}},off(c,p){var d,u,_,f;switch(c.type){case Ws.Event:if(p!==void 0){(d=r.get(c.id))==null||d.delete(p),(u=t.get(c.id))==null||u.delete(p);break}r.delete(c.id),t.delete(c.id);break;case Ws.MatchExpression:if(p!==void 0){(_=o.get(c.id))==null||_.delete(p),(f=n.get(c.id))==null||f.delete(p);break}o.delete(c.id),n.delete(c.id);break}}}}let tn=function(e){return e[e.SendEvent=0]="SendEvent",e[e.SendEventError=1]="SendEventError",e[e.ReceiveEvent=2]="ReceiveEvent",e[e.ReceiveEventError=3]="ReceiveEventError",e[e.ReceiveEventStreamEnd=4]="ReceiveEventStreamEnd",e}({});function Pw(e){e||(e=So());const r={...cs(`${e}-send`),invokeType:tn.SendEvent},t={...cs(`${e}-send-error`),invokeType:tn.SendEventError},s={...cs(`${e}-receive`),invokeType:tn.ReceiveEvent},o={...cs(`${e}-receive-error`),invokeType:tn.ReceiveEventError},n={...cs(`${e}-receive-stream-end`),invokeType:tn.ReceiveEventStreamEnd};return{sendEvent:r,sendEventError:t,receiveEvent:s,receiveEventError:o,receiveEventStreamEnd:n}}function iE(e){return typeof e!="object"?!1:"invokeType"in e}function lE(e){return iE(e)?e.invokeType===tn.ReceiveEvent||e.invokeType===tn.ReceiveEventError||e.invokeType===tn.ReceiveEventStreamEnd:!1}function uE(e){var r,t;return lE(e)?typeof((r=e.body)==null?void 0:r.content)=="object"&&((t=e.body)==null?void 0:t.content)!=null&&"response"in e.body.content&&(!("invokeResponse"in e.body.content)||"invokeResponse"in e.body.content&&(typeof e.body.content.invokeResponse=="object"||typeof e.body.content.invokeResponse>"u")):!1}function Cw(e,r,t){var n,a;e.invokeHandlers||(e.invokeHandlers=new Map);let s=(n=e.invokeHandlers)==null?void 0:n.get(r.sendEvent.id);s||(s=new Map,(a=e.invokeHandlers)==null||a.set(r.sendEvent.id,s));let o=s.get(t);return o||(o=async(i,l)=>{var c;if(i.body&&i.body.invokeId)try{const p=await t((c=i.body)==null?void 0:c.content,l);e.emit({...cs(`${r.receiveEvent.id}-${i.body.invokeId}`),invokeType:r.receiveEvent.invokeType},{...i.body,content:p},l)}catch(p){e.emit({...cs(`${r.receiveEventError.id}-${i.body.invokeId}`),invokeType:r.receiveEventError.invokeType},{...i.body,content:p},l)}},s.set(t,o),e.on(r.sendEvent,o)),()=>e.off(r.sendEvent,o)}function cE(e,r){return{id:So(),type:e,payload:r}}function jE(e){return e}function Sw(e){return typeof e=="object"&&"_workerTransfer"in e&&typeof e._workerTransfer=="boolean"&&e._workerTransfer===!0}const $w=cs();cs();function dE(e,r){var o,n,a,i,l;let t=e.body,s;return uE(e)?(((o=e.body.content.invokeResponse)==null?void 0:o.transfer)!=null&&(s=e.body.content.invokeResponse.transfer,delete e.body.content.invokeResponse),t={...e.body,content:e.body.content.response},delete t.content.response):Sw(e)&&(s=(n=e.body)==null?void 0:n.transfer,(a=e.body)==null||delete a.transfer,t=(i=e.body)==null?void 0:i.message,(l=e.body)==null||delete l.message),typeof r<"u"&&r!=null&&typeof r=="object"&&"transfer"in r&&Array.isArray(r.transfer)&&(s=r.transfer),{body:t,transfer:s}}function pE(e){const{messagePort:r=self}={},t=aE();return t.on(oE(Ew(s=>s._flowDirection===Pu.Outbound||!s._flowDirection),Ew("*")),(s,o)=>{const{body:n,transfer:a}=dE(s,o),i=cE(s.id,{...nE(s.type),...s,body:n});if(a!=null){r.postMessage(i,{transfer:a});return}r.postMessage(i)}),self.onerror=s=>{t.emit($w,{error:s},{raw:{error:s}})},self.onmessage=s=>{try{const{type:o,payload:n}=s.data;Sw(n)?t.emit(Tw(o),{message:n.body},{raw:{event:s}}):t.emit(Tw(o),n.body,{raw:{event:s}})}catch(o){console.error("Failed to parse WebWorker message:",o),t.emit($w,{error:o},{raw:{event:s}})}},{context:t}}const hE=Pw("vlm-player:eventa:invoke:load-model"),mE=Pw("vlm-player:eventa:invoke:generate"),fE=cs("vlm-player:eventa:event:model-load-progress"),Aw="onnx-community/FastVLM-0.5B-ONNX";let Cu=null,$o=null;const{context:Su}=pE();async function _E(){const e={};let r=0;Cu=await GT.from_pretrained(Aw,{dtype:{embed_tokens:"fp16",vision_encocder:"q4",decoder_model_merged:"q4"},progress_callback:t=>{if(t.status!=="progress")return;e[t.name]=t.progress;const s=Object.values(e).reduce((o,n)=>o+n,0)/Object.values(e).length;s>r&&(r=s,Su.emit(fE,r))},device:"webgpu"}),$o=await HT.from_pretrained(Aw)}async function gE(e){if(!$o||!Cu)throw new Error("Model or processor not loaded");const r=await KT.fromBlob(e),t=[{role:"rule",content:"<image>You are playing the game Factorio, you will move the character based on the image. You MUST only use one of the keycodes W(up), A(left), S(down), or D(right) in the output."}],s=$o.apply_chat_template(t,{add_generation_prompt:!0}),o=await $o(r,s,{add_special_tokens:!1}),n=await Cu.generate({...o,max_new_tokens:256,do_sample:!1});return $o.batch_decode(n.slice(null,[o.input_ids.dims.at(-1),null]),{skip_special_tokens:!0})[0]}function kw(e){return async(...r)=>{try{return await e(...r)}catch(t){throw console.error(t),t}}}Cw(Su,hE,kw(_E)),Cw(Su,mE,kw(gE))})();
//# sourceMappingURL=vlm-play-worker-CFOrl6pM.js.map