Update app.py

app.py

@@ -7,8 +7,7 @@ MODEL_NAME = "beyoru/Qwen3-0.9B-A0.6B"
 tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
 model = AutoModelForCausalLM.from_pretrained(
     MODEL_NAME,
-    torch_dtype=torch.bfloat16,
-    device_map="auto"
+    torch_dtype=torch.float16,
 )
 
 # --- Chat function ---
@@ -30,7 +29,7 @@ def chat_fn(message, history, num_ctx, temperature, repeat_penalty, min_p, top_k
 
     outputs = model.generate(
         **inputs,
-        max_new_tokens=4096,
+        max_new_tokens=2048,
         temperature=float(temperature),
         top_p=float(top_p),
         top_k=int(top_k),
@@ -51,7 +50,7 @@ with gr.Blocks(fill_height=True, fill_width=True) as app:
         gr.Markdown("## Qwen3 Playground (Transformers Edition)")
         gr.Markdown("Model: **beyoru/Qwen3-0.9B-A0.6B** — chạy trực tiếp bằng Transformers")
 
-        num_ctx = gr.Slider(512, 8192, 8192, 128, label="Context Length (num_ctx)")
+        num_ctx = gr.Slider(512, 8192, 2048, 128, label="Context Length (num_ctx)")
         temperature = gr.Slider(0.1, 2.0, 0.6, 0.1, label="Temperature")
         repeat_penalty = gr.Slider(0.1, 2.0, 1.0, 0.1, label="Repeat Penalty")
         min_p = gr.Slider(0.0, 1.0, 0.0, 0.01, label="Min P")