Update app.py

app.py

@@ -2,7 +2,7 @@ from threading import Thread
 from typing import Dict
 
 import gradio as gr
-import spaces
+# import spaces
 import torch
 from PIL import Image
 from transformers import AutoModelForVision2Seq, AutoProcessor, AutoTokenizer, TextIteratorStreamer
@@ -28,7 +28,7 @@ processor = AutoProcessor.from_pretrained(model_id)
 model = AutoModelForVision2Seq.from_pretrained(model_id, torch_dtype="auto", device_map="auto")
 
 
-@spaces.GPU
+# @spaces.GPU
 def stream_chat(message: Dict[str, str], history: list):
     # Turn 1:
     # {'text': 'what is this', 'files': ['image-xxx.jpg']}
@@ -39,8 +39,9 @@ def stream_chat(message: Dict[str, str], history: list):
     # [[('image-xxx.jpg',), None], ['what is this', 'a image.']]
 
     image_path = None
-    if len(message["files"]) != 0:
-        image_path = message["files"][0]
+    # print(message.files[0].path)
+    if len(message.files) != 0:
+        image_path = message.files[0].path
 
     if len(history) != 0 and isinstance(history[0][0], tuple):
         image_path = history[0][0][0]
@@ -57,7 +58,7 @@ def stream_chat(message: Dict[str, str], history: list):
     for prompt, answer in history:
         conversation.extend([{"role": "user", "content": prompt}, {"role": "assistant", "content": answer}])
 
-    conversation.append({"role": "user", "content": message["text"]})
+    conversation.append({"role": "user", "content": message.text})
 
     input_ids = tokenizer.apply_chat_template(conversation, add_generation_prompt=True, return_tensors="pt")
     image_token_id = tokenizer.convert_tokens_to_ids("<image>")
@@ -100,3 +101,4 @@ with gr.Blocks(css=CSS) as demo:
 
 if __name__ == "__main__":
     demo.launch()
+