fix app

app.py

@@ -1,5 +1,5 @@
 import gradio as gr
-from transformers import AutoModelForVision2Seq, AutoProcessor, AutoModelForVision2Seq, BitsAndBytesConfig
+from transformers import AutoModelForVision2Seq, AutoProcessor, BitsAndBytesConfig
 import torch
 
 
@@ -30,9 +30,15 @@ def respond(multimodal_input):
     generated_text = processor.batch_decode(new_tokens, skip_special_tokens=True)[0]
     return generated_text
 
-title = "IDEFICS2-8B DPO"
-description = "Try IDEFICS2-8B fine-tuned using direct preference optimization (DPO) in this demo. Learn more about vision language model DPO integration of TRL [here]."
-examples=[{"text": "What is the type of flower in the image and what insect is on it?", "files": ["./bee.jpg"]},
-             {"text": "Describe the image", "files": ["./howl.jpg"]}],
-gr.Interface(respond, inputs=[gr.MultimodalTextbox(file_types=["image"], show_label=False)], outputs="text",
-            title=title, description=description, examples=examples).launch()
+
+gr.Interface(
+    respond,
+    inputs=[gr.MultimodalTextbox(file_types=["image"], show_label=False)],
+    outputs="text",
+    title="IDEFICS2-8B DPO",
+    description="Try IDEFICS2-8B fine-tuned using direct preference optimization (DPO) in this demo. Learn more about vision language model DPO integration of TRL [here](https://huggingface.co/blog/dpo_vlm).",
+    examples=[
+        {"text": "What is the type of flower in the image and what insect is on it?", "files": ["./bee.jpg"]},
+        {"text": "Describe the image", "files": ["./howl.jpg"]},
+    ],
+).launch()