Update app.py

app.py

@@ -1,6 +1,4 @@
 import subprocess
-# Installing flash_attn
-subprocess.run('pip install flash-attn --no-build-isolation', env={'FLASH_ATTENTION_SKIP_CUDA_BUILD': "TRUE"}, shell=True)
 
 import gradio as gr
 from PIL import Image 
@@ -13,9 +11,9 @@ import torch
 import spaces
 
 model_id = "microsoft/Phi-3-vision-128k-instruct" 
-model = AutoModelForCausalLM.from_pretrained(model_id, device_map="cuda", trust_remote_code=True, torch_dtype="auto")
+model = AutoModelForCausalLM.from_pretrained(model_id, device_map="cpu", trust_remote_code=True, torch_dtype="auto")
 processor = AutoProcessor.from_pretrained(model_id, trust_remote_code=True) 
-model.to("cuda:0")
+model.to("cpu")
 
 PLACEHOLDER = """
 <div style="padding: 30px; text-align: center; display: flex; flex-direction: column; align-items: center;">
@@ -71,7 +69,7 @@ def bot_streaming(message, history):
     print(f"prompt is -\n{conversation}")
     prompt = processor.tokenizer.apply_chat_template(conversation, tokenize=False, add_generation_prompt=True)
     image = Image.open(image)
-    inputs = processor(prompt, image, return_tensors="pt").to("cuda:0") 
+    inputs = processor(prompt, image, return_tensors="pt").to("cpu") 
 
     streamer = TextIteratorStreamer(processor, **{"skip_special_tokens": True, "skip_prompt": True, 'clean_up_tokenization_spaces':False,}) 
     generation_kwargs = dict(inputs, streamer=streamer, max_new_tokens=1024, do_sample=False, temperature=0.0, eos_token_id=processor.tokenizer.eos_token_id,)