search

facebookresearch / ToMe

A method to increase the speed and lower the memory footprint of existing vision transformers.
Other
970 stars 69 forks source link

Apply Flash-attn for MAE-based models. #33

Open JerryFlymi opened 1 year ago

JerryFlymi commented 1 year ago

According to the paper： "Surprisingly, proportional attention is necessary for supervised models (e.g., AugReg, SWAG, DeiT), but not for MAE models." We can easily arm MAE-based models with Flash attention.

facebook-github-bot commented 1 year ago

Hi @JerryFlymi!

Thank you for your pull request and welcome to our community.

Action Required

In order to merge any pull request (code, docs, etc.), we require contributors to sign our Contributor License Agreement, and we don't seem to have one on file for you.

Process

In order for us to review and merge your suggested changes, please sign at https://code.facebook.com/cla. If you are contributing on behalf of someone else (eg your employer), the individual CLA may not be sufficient and your employer may need to sign the corporate CLA.

Once the CLA is signed, our tooling will perform checks and validations. Afterwards, the pull request will be tagged with CLA signed. The tagging process may take up to 1 hour after signing. Please give it that time before contacting us about it.

If you have received this in error or have any questions, please contact us at cla@meta.com. Thanks!

facebook-github-bot commented 1 year ago

Thank you for signing our Contributor License Agreement. We can now accept your code for this (and any) Meta Open Source project. Thanks!

dbolya commented 1 year ago

Good idea! I don't want to require a new library for ToMe though and I'd rather flash attention be default off to be consistent with the paper.

As for requiring a library, have you tried torch.nn.functional's scaled_dot_product_attention? Is the performance similar to flash attention? Because if so, we should probably use that instead.