Update app.py

app.py

@@ -4,7 +4,7 @@ from transformers import pipeline, TextIteratorStreamer
 import torch
 import threading
 
-# Load model pipeline
+# Load pipeline and components
 model_name = "krish10/Qwen3_0.6B_16bit_TA_screen"
 pipe = pipeline("text-generation", model=model_name, device=0)
 tokenizer = pipe.tokenizer
@@ -15,7 +15,7 @@ MAX_TOKENS = 3000
 TEMPERATURE = 0.1
 TOP_P = 0.9
 
-# Response function using a streamer
+# Generator function for streaming
 @spaces.GPU
 def respond_stream(user_input):
     messages = [{"role": "user", "content": user_input}]
@@ -37,30 +37,31 @@ def respond_stream(user_input):
     thread = threading.Thread(target=model.generate, kwargs=generation_kwargs)
     thread.start()
 
-    # Yield tokens as they come
-    for token in streamer:
-        yield token
+    partial_text = ""
+    for new_token in streamer:
+        partial_text += new_token
+        yield partial_text
 
-# Gradio UI with streaming
+# Gradio interface with live streaming (no stream=True)
 with gr.Blocks() as demo:
-    gr.Markdown("## 🤖 Qwen Chat with Streaming (Fixed Temp/Top-p/Max Tokens)")
+    gr.Markdown("## 🤖 Qwen Streaming Chat (Fixed Parameters)")
 
     input_box = gr.Textbox(
         lines=12,
         label="Your input (instruction + abstract)",
         placeholder="Instruction: Population = ...\nAbstract: ..."
     )
-    output_box = gr.Textbox(lines=12, label="Model Response")
+
+    output_box = gr.Textbox(lines=12, label="Model Response", interactive=False)
+
     generate_btn = gr.Button("Generate")
 
     generate_btn.click(
         fn=respond_stream,
         inputs=[input_box],
-        outputs=[output_box],
-        api_name="chat",
-        stream=True
+        outputs=[output_box]
     )
 
-# Launch app
+# Launch
 if __name__ == "__main__":
     demo.launch()