Update app.py

app.py

@@ -59,12 +59,13 @@ def stream_chat(message, history: list, temperature: float, max_new_tokens: int)
         conversation.append({"role": "user", "content": f"<|image_1|>\n{message['text']}"})
     else:
         if len(history) == 0:
-            gr.Error("Please upload an image first.")
+            raise gr.Error("Please upload an image first.")
         image = None
         conversation.append({"role": "user", "content": message['text']})
-       
-    prompt = processor.tokenizer.apply_chat_template(conversation, tokenize=False, add_generation_prompt=True)
-    inputs = processor(prompt, images=image, return_tensors="pt").to(0)
+    print(conversation)
+    
+    inputs = processor.tokenizer.apply_chat_template(conversation, tokenize=False, add_generation_prompt=True)
+    inputs_ids = processor(inputs, image, return_tensors="pt").to(0)
     
     generate_kwargs = dict(
         max_new_tokens=max_new_tokens,
@@ -74,7 +75,7 @@ def stream_chat(message, history: list, temperature: float, max_new_tokens: int)
     )
     if temperature == 0:
         generate_kwargs["do_sample"] = False
-    generate_kwargs = {**inputs, **generate_kwargs}
+    generate_kwargs = {**inputs_ids, **generate_kwargs}
     
     streamer = TextIteratorStreamer(processor, **{"skip_special_tokens": True, "skip_prompt": True, 'clean_up_tokenization_spaces':False,})