Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

#9336: Refactoring moreh layernorm #9636

Merged
merged 18 commits into from
Jul 5, 2024

Conversation

hschoi4448
Copy link
Contributor

@hschoi4448 hschoi4448 commented Jun 24, 2024

Ticket

Problem description

  • The current moreh_layernorm only accepts 4D tensors as input. To meet this restriction, the mean and rstd tensors are stored inefficiently, with only one valid data point per tile.
  • the current implementation does not support optional outputs and the fp32_dest_acc_en feature.

What's changed

  • refactoring moreh_layernorm
    • add fp32_dest_acc_en support
    • support non 4d tensor
    • support optional output tensor

Checklist

  • Post commit CI passes
  • Model regression CI testing passes (if applicable)
  • New/Existing tests provide coverage for changes

@hschoi4448 hschoi4448 added the moreh moreh contribution label Jun 24, 2024
@hschoi4448 hschoi4448 force-pushed the hyungsuk/refactoring_moreh_layernorm branch from 7982b66 to efc633d Compare July 5, 2024 01:40
@hschoi4448 hschoi4448 merged commit c52e153 into main Jul 5, 2024
80 checks passed
@hschoi4448 hschoi4448 deleted the hyungsuk/refactoring_moreh_layernorm branch July 5, 2024 02:51
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
moreh moreh contribution
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

4 participants