Update app.py

app.py

@@ -53,7 +53,7 @@ def ovis_chat(chatbot, image_input):
         "value": text_input
     })
     if image_input is not None:
-        conversations[0]["value"] = image_placeholder + '\n' + conversations[0]["value"]
+        conversations[1]["value"] = image_placeholder + '\n' + conversations[1]["value"]
     prompt, input_ids, pixel_values = model.preprocess_inputs(conversations, [image_input])
     attention_mask = torch.ne(input_ids, text_tokenizer.pad_token_id)
     input_ids = input_ids.unsqueeze(0).to(device=model.device)
@@ -76,18 +76,25 @@ def ovis_chat(chatbot, image_input):
             use_cache=True
         )
         response = ""
-        thread = Thread(target=model.generate, 
-                    kwargs={"inputs": input_ids,
-                            "pixel_values": pixel_values,
-                            "attention_mask": attention_mask,
-                            "streamer": streamer,
-                            **gen_kwargs})
-        thread.start()
+        # thread = Thread(target=model.generate, 
+        #             kwargs={"inputs": input_ids,
+        #                     "pixel_values": pixel_values,
+        #                     "attention_mask": attention_mask,
+        #                     "streamer": streamer,
+        #                     **gen_kwargs})
+        model.generate(
+                input_ids,
+                pixel_values=pixel_values,
+                attention_mask=attention_mask,
+                streamer=streamer,
+                **gen_kwargs
+            )
+        # thread.start()
         for new_text in streamer:
             response += new_text
             chatbot[-1][1] = response
             yield chatbot
-        thread.join()
+        # thread.join()
         # debug
         print('*'*60)
         print('*'*60)