update

app.py

@@ -12,11 +12,22 @@ model = Qwen2_5_VLForConditionalGeneration.from_pretrained(
 )
 
 @spaces.GPU
-def respond(input_dict, history):
+def respond(
+    input_dict,
+    chat_history,
+    system_message,
+    max_tokens,
+    temperature,
+    top_p,
+):
     text = input_dict["text"]
     files = input_dict["files"]
 
-    messages = [
+    messages = [{
+        "role": "system",
+        "content": system_message
+    }]
+    messages.append(
         {
             "role": "user",
             "content": [
@@ -24,7 +35,7 @@ def respond(input_dict, history):
                 *[{"type": "image", "image": image} for image in files]
             ]
         }
-    ]
+    )
     image_inputs, video_inputs = process_vision_info(messages)
     prompt = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
     inputs = processor(
@@ -36,7 +47,7 @@ def respond(input_dict, history):
     ).to(model.device)
     
     streamer = TextIteratorStreamer(processor, skip_prompt=True, skip_special_tokens=True)
-    generation_kwargs = dict(inputs, streamer=streamer, max_new_tokens=1024)
+    generation_kwargs = dict(inputs, streamer=streamer, max_new_tokens=max_tokens, temperature=temperature, top_p=top_p)
 
     thread = Thread(target=model.generate, kwargs=generation_kwargs)
     thread.start()
@@ -50,6 +61,18 @@ demo = gr.ChatInterface(
         fn=respond,
         type='messages',
         multimodal=True,
+        additional_inputs=[
+            gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
+            gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
+            gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
+            gr.Slider(
+                minimum=0.1,
+                maximum=1.0,
+                value=0.95,
+                step=0.05,
+                label="Top-p (nucleus sampling)",
+            ),
+        ],
     )
 
 demo.launch(debug=True)