fix triton attention bug

[zhanglei1172]

The implementation of Triton does not carry out log2 scaling for bias, resulting in a large error. After correction, I can obtain an error with an absolute value <0.015 in my test here.

[facebook-github-bot]

Hi @zhanglei1172!

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 [email protected]. Thanks!

[cpuhrsch]

Thank you for the PR @zhanglei1172! Can you also add the test that you used to build this?

Just to make sure you won't think I forgot about this, I'm going to be on vacation until November 27th starting today. I'll review it again once I'm back! Thank you again for sending a PR!

[cpuhrsch]

Hey @zhanglei1172 - I merged this as part of #108 . I was going to push to this pull request, but didn't want to push to your fork's main branch and interfere with something there. Thank yo for filing this PR and fixing the issue!