Update app.py

app.py

@@ -3,20 +3,19 @@ import torch
 import gradio as gr
 from transformers import AutoModelForCausalLM, AutoProcessor
 
-# Clone the model if not already present
+# Clone the model if not already downloaded
 if not os.path.exists("VideoLLaMA3-7B"):
     os.system("apt-get update && apt-get install -y git git-lfs && git lfs install")
     os.system("git clone https://huggingface.co/DAMO-NLP-SG/VideoLLaMA3-7B")
 
-# Load model and processor from the local clone
 model_path = "./VideoLLaMA3-7B"
 
+# Load model (no flash_attn)
 model = AutoModelForCausalLM.from_pretrained(
     model_path,
     trust_remote_code=True,
     device_map="auto",
     torch_dtype=torch.bfloat16,
-    attn_implementation="flash_attention_2",
 )
 processor = AutoProcessor.from_pretrained(model_path, trust_remote_code=True)
 
@@ -38,7 +37,7 @@ def describe_video(video, question):
     output_ids = model.generate(**inputs, max_new_tokens=128)
     return processor.batch_decode(output_ids, skip_special_tokens=True)[0].strip()
 
-# Gradio UI
+# Gradio interface
 demo = gr.Interface(
     fn=describe_video,
     inputs=[