某二次元的高坂桐乃

About the csv file

3

simclr/all_patches.csv How to generate the csv file? And what kind of data is stored in it?

Thanks for your wonderful work. I wonder if using weight provided by Video-MAE directly for weight initialization will lead to unfair comparison of pre-training tasks?

load_from_name 加入 flash-attn 支持

4

感谢你如此好的代码实现，他对我的帮助很大，但是我在使用load_from_name 函数时，我发现并不支持flash-attn ，因此我自己实现了这一块的代码，但是我不确定实现是否正确，尽管它可以正常运行。 以下是代码片段 ``` ###### ------- ps: add use_flash_attention keyword ------- ###### def load_from_name(name: str, device: Union[str, torch.device] = "cuda" if torch.cuda.is_available() else "cpu", download_root: str = None, vision_model_name:...

Thank you for your exciting work. I would like to research this direction further because this architecture is simple and effective. Could you share the pre-trained weights for Resnet18 and...