test

app.py

@@ -16,7 +16,6 @@ auth_token = os.environ.get("TOKEN_FROM_SECRET")
 import torch
 from transformers import AutoProcessor, AutoTokenizer
 from transformers import Qwen2ForCausalLM, Qwen2_5_VLForConditionalGeneration, TextIteratorStreamer
-from vllm import LLM, SamplingParams
 from qwen_vl_utils import process_vision_info
 from threading import Thread
 
@@ -108,7 +107,7 @@ def add_text(state, text, image, image_process_mode):
     logging.info(f"add_text. len: {len(text)}")
     if len(text) <= 0 or image is None:
         state.skip_next = True
-        return (state, state.to_gradio_chatbot_public(), "", None) + (no_change_btn,) * 2
+        return (state, state.to_gradio_chatbot_public(), "") + (no_change_btn,) * 2
     
     # Deal with image inputs
     if image is not None:
@@ -140,6 +139,8 @@ def http_bot(state):
     pload = {"prompt": prompt, "images": f'List of {len(state.get_images())} images: {all_images}'}
     logging.info(f"==== request ====\n{pload}")
     
+    return
+    
     # Construct prompt
     cap_msgs, qa_msgs = build_messages(all_images, prompt)
     cap_prompt = processor.apply_chat_template([cap_msgs], tokenize=False, add_generation_prompt=True)
@@ -220,7 +221,7 @@ def build_demo(embed_mode):
                     label="Preprocess for non-square image", visible=False)
 
                 gr.Examples(examples=[
-                    ["./examples/demo_example.png", "When the canister is momentarily stopped by the spring, by what distance $d$ is the spring compressed?"],
+                    ["./examples/image-text/demo_example.png", "When the canister is momentarily stopped by the spring, by what distance $d$ is the spring compressed?"],
                 ], inputs=[imagebox, textbox], label='Examples')
 
             with gr.Column(scale=8):