Update app.py

app.py

@@ -76,26 +76,26 @@ def bot_streaming(message, history, temperature, top_k, max_new_tokens):
     }
     
     # Generate from the VLM
-    buffer = vlm.generate(
-    **generation_kwargs
-    )
-    # thread = Thread(target=vlm.generate, kwargs=generation_kwargs)
-    # thread.start()
-
-    # buffer = ""
-    # output_started = False
-    # for new_text in streamer:
-    #     if not output_started:
-    #         if "<|assistant|>\n" in new_text:
-    #             output_started = True
-    #         continue
-    #     buffer += new_text
-    #     if len(buffer) > 1:
-    #         yield buffer
+    # buffer = vlm.generate(
+    # **generation_kwargs
+    # )
+    thread = Thread(target=vlm.generate, kwargs=generation_kwargs)
+    thread.start()
+
+    buffer = ""
+    output_started = False
+    for new_text in streamer:
+        if not output_started:
+            if "<|assistant|>\n" in new_text:
+                output_started = True
+            continue
+        buffer += new_text
+        if len(buffer) > 1:
+            yield buffer
 
     prompt_builder.add_turn(role="gpt", message=buffer)
 
-    yield buffer
+    return buffer
     
     
 demo = gr.ChatInterface(fn=bot_streaming,