New README #392
New README #392
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/392
Note: Links to docs will display an error until the docs builds have been completed. ⏳ No Failures, 1 PendingAs of commit 7464e55 with merge base aeee551 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
facebook-github-bot
added
the
CLA Signed
README.md
Outdated
|
|
||
| ### Training | ||
|
|
||
| We've added support for semi-structured 2:4 sparsity with over 30% speedups on ViT-L |
There was a problem hiding this comment.
to clarify, the speedup here is not for the entire model. I believe its specifically for the MLP blocks in ViT. cc @jcaip
There was a problem hiding this comment.
can we put together anything from an end to end perspective
README.md
Outdated
|
|
||
| For VIT-L MLP shapes on a NVIDIA A100 we see the following results: | ||
|
|
||
| | | act24 | dense | w24 | s24_inp_sparsify24 | s24_inp_clone | |
There was a problem hiding this comment.
discussed this with @jcaip offline but would be good to make the columns here more clear.
There was a problem hiding this comment.
@jcaip please send me a clearer table when you have it - keep in mind I want to make some claim about end to end performance
There was a problem hiding this comment.
Yes, I'm making some changes to the benchmarking script for e2e ViT. Will send something by EOD.
* New README * yolo * yolo * Update README.md * Update README.md * Trigger CI * Trigger CI * Trigger CI * push * push * push * push * push * push * push * push * push * push * push
This a restructuring of our main README.md with the main goals of having