Update app.py

app.py

@@ -86,6 +86,8 @@ print(f"Gradio-client version: {gradio_client.__version__}")
 def get_conv_log_filename():
     t = datetime.datetime.now()
     name = os.path.join(LOGDIR, f"{t.year}-{t.month:02d}-{t.day:02d}-user_conv.json")
+    if not os.path.isfile(name):
+        os.makedirs(os.path.dirname(name), exist_ok=True)
     return name
 
 def get_conv_vote_filename():
@@ -249,10 +251,13 @@ def clear_history(history):
 
 def add_message(history, message):
     global chat_image_num
+    print("#### len(history)",len(history))
     if not history:
         history = []
+        print("### Initialize chatbot")
         our_chatbot = chat_manager.get_chatbot(args, model_path, tokenizer, model, image_processor, context_len)
         chat_image_num = 0
+        print("chat_image_num", chat_image_num)
         
     if len(message["files"]) <= 1:
         for x in message["files"]:
@@ -269,8 +274,8 @@ def add_message(history, message):
                 
         if message["text"] is not None:
             history.append((message["text"], None))
-        
-        print(f"### Chatbot instance ID: {id(our_chatbot)}")
+        print("chat_image_num", chat_image_num)
+        # print(f"### Chatbot instance ID: {id(our_chatbot)}")
         return history, gr.MultimodalTextbox(value=None, interactive=False)
     else:
         for x in message["files"]:
@@ -341,22 +346,26 @@ def bot(history, temperature, top_p, max_output_tokens):
             image_list.append(load_image(f))
         else:
             raise ValueError("Invalid image file")
-    
-    image_tensor = [
-        our_chatbot.image_processor.preprocess(f, return_tensors="pt")["pixel_values"][
-            0
+            
+    image_tensor = []
+    if num_new_images > 0:
+        image_tensor = [
+            our_chatbot.image_processor.preprocess(f, return_tensors="pt")["pixel_values"][
+                0
+            ]
+            .half()
+            .to(our_chatbot.model.device)
+            for f in image_list
         ]
-        .half()
-        .to(our_chatbot.model.device)
-        for f in image_list
-    ]
-
-
-    image_tensor = torch.stack(image_tensor)
-    image_token = DEFAULT_IMAGE_TOKEN * num_new_images
     
-    inp = text
-    inp = image_token + "\n" + inp
+    
+        image_tensor = torch.stack(image_tensor)
+        image_token = DEFAULT_IMAGE_TOKEN * num_new_images
+    
+        inp = text
+        inp = image_token + "\n" + inp
+    else:
+        inp = text
     our_chatbot.conversation.append_message(our_chatbot.conversation.roles[0], inp)
     # image = None
     our_chatbot.conversation.append_message(our_chatbot.conversation.roles[1], None)
@@ -381,12 +390,12 @@ def bot(history, temperature, top_p, max_output_tokens):
     )
     print(our_chatbot.model.device)
     print(input_ids.device)
-    print(image_tensor.device)
+    # print(image_tensor.device)
     
     generate_kwargs = dict(
         inputs=input_ids,
         streamer=streamer,
-        images=image_tensor,
+        images=image_tensor if num_new_images > 0 else None,
         do_sample=True,
         temperature=temperature,
         top_p=top_p,