1. 18 Feb, 2022 1 commit
  2. 16 Feb, 2022 1 commit
    • Sangkug Lym's avatar
      gradient accumulation fusion · 83b1e42f
      Sangkug Lym authored
      remove redundant linear layer class definition
      
      add fuse_gradient_accumulation attribute to weights for simple targetting
      
      reflect feedback and clean up the codes
      
      arg change
      83b1e42f
  3. 15 Feb, 2022 4 commits
  4. 11 Feb, 2022 1 commit
  5. 04 Feb, 2022 1 commit
  6. 03 Feb, 2022 1 commit
  7. 01 Feb, 2022 5 commits
  8. 31 Jan, 2022 1 commit
  9. 29 Jan, 2022 1 commit
  10. 27 Jan, 2022 1 commit
  11. 26 Jan, 2022 1 commit
  12. 25 Jan, 2022 1 commit
  13. 24 Jan, 2022 2 commits
  14. 22 Jan, 2022 1 commit
  15. 13 Jan, 2022 1 commit
  16. 12 Jan, 2022 1 commit
  17. 11 Jan, 2022 2 commits
  18. 10 Jan, 2022 1 commit
  19. 08 Jan, 2022 1 commit
  20. 07 Jan, 2022 1 commit
  21. 04 Jan, 2022 1 commit
  22. 31 Dec, 2021 1 commit
  23. 30 Dec, 2021 2 commits
  24. 17 Dec, 2021 3 commits
  25. 10 Dec, 2021 1 commit
  26. 22 Nov, 2021 1 commit
  27. 18 Nov, 2021 1 commit
  28. 11 Nov, 2021 1 commit
    • Sangkug Lym's avatar
      persistent fused layer norm · a2fdcdf0
      Sangkug Lym authored
      fix the guard to fall back to the baseline fused layer norm kernel
      
      Persisten ln: move the guard for supported hidden sizes to layer norm module
      a2fdcdf0