update

app.py

@@ -12,7 +12,7 @@ processor = AutoProcessor.from_pretrained("MFuyu/mantis-8b-idefics2-video-eval_8
 model = Idefics2ForConditionalGeneration.from_pretrained("MFuyu/mantis-8b-idefics2-video-eval_8192_lora")
 conv_template = conv_templates["idefics_2"]
 
-with open("./examples/data_subset.json", 'r') as f:
+with open("./examples/all_subsets.json", 'r') as f:
     examples = json.load(f)
 
 for item in examples:
@@ -153,7 +153,8 @@ def bot(history):
     }
     
     response = generate(None, chat_images, chat_history, **generation_kwargs) 
-    return response
+    history[-1][1] = response
+    return history
     # for _output in response:
     #     history[-1][1] = _output
     #     time.sleep(0.05)
@@ -183,7 +184,6 @@ Mantis is a multimodal conversational AI model that can chat with users about im
         gr.Markdown("""## Chat with Mantis
         Mantis supports interleaved text-image input format, where you can simply use the placeholder `<image>` to indicate the position of uploaded images.
         The model is optimized for multi-image reasoning, while preserving the ability to chat about text and images in a single conversation.
-        (The model currently serving is [🤗 TIGER-Lab/Mantis-8B-siglip-llama3](https://huggingface.co/TIGER-Lab/Mantis-8B-siglip-llama3))
         """)
         
         chatbot = gr.Chatbot(line_breaks=True)