John Ho commited on
Commit
f87fafd
·
1 Parent(s): 2a9891d

try use flash attention again

Browse files
Files changed (1) hide show
  1. app.py +1 -1
app.py CHANGED
@@ -69,7 +69,7 @@ def inference(
69
  ):
70
  # default processor
71
  processor = AutoProcessor.from_pretrained("Qwen/Qwen2.5-VL-7B-Instruct")
72
- model = load_model(use_flash_attention=False)
73
  fps = get_fps_ffmpeg(video_path)
74
  logger.info(f"{os.path.basename(video_path)} FPS: {fps}")
75
  messages = [
 
69
  ):
70
  # default processor
71
  processor = AutoProcessor.from_pretrained("Qwen/Qwen2.5-VL-7B-Instruct")
72
+ model = load_model(use_flash_attention=True)
73
  fps = get_fps_ffmpeg(video_path)
74
  logger.info(f"{os.path.basename(video_path)} FPS: {fps}")
75
  messages = [