V100 not support FlashAttention，how to disable FlashAttention function？

[mishaogui]

[mishaogui]

run demo.py python demo/demo.py ./images --model_path ./Sa2VA-4B --work-dir ./OUTPUT_DIR --text "Please describe the video content."

[Subcode]

model = AutoModel.from_pretrained(
model_path,
torch_dtype=torch.bfloat16,
low_cpu_mem_usage=True,
use_flash_attn=False,
trust_remote_code=True,
).eval().cuda()

in projects/llava_sam2/gradio/app.py

for demo.py add use_flash_attn=False to AutoModelFromCausalLLM.from_pretrained().