Skip to content

Conversation

@ptillet
Copy link
Collaborator

@ptillet ptillet commented Jun 3, 2025

  • rename bitmatrix.py -> datastruct.py
  • added option to compute softmax before topk
  • added backward pass support to routing
  • refactor expt_data to precompute infos for all possible block sizes
  • added support for dynamic batch size + padded inputs
  • sort topk output in order of increasing indices

@ptillet ptillet marked this pull request as ready for review June 5, 2025 02:35
@apgoucher apgoucher merged commit b57ff70 into main Jun 5, 2025
8 checks passed
@apgoucher apgoucher deleted the phil/kernels/routing-update branch June 5, 2025 23:43
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants