1. 01 Mar, 2022 1 commit
  2. 23 Feb, 2022 1 commit
  3. 22 Feb, 2022 1 commit
  4. 15 Feb, 2022 1 commit
  5. 10 Feb, 2022 2 commits
  6. 28 Jan, 2022 2 commits
  7. 27 Jan, 2022 1 commit
  8. 26 Jan, 2022 1 commit
  9. 22 Jan, 2022 1 commit
  10. 19 Jan, 2022 1 commit
  11. 12 Jan, 2022 1 commit
  12. 21 Dec, 2021 1 commit
  13. 17 Dec, 2021 1 commit
  14. 02 Dec, 2021 1 commit
  15. 26 Nov, 2021 1 commit
  16. 24 Nov, 2021 1 commit
  17. 01 Oct, 2021 1 commit
  18. 20 Sep, 2021 1 commit
  19. 19 Aug, 2021 1 commit
  20. 18 Aug, 2021 2 commits
  21. 17 Aug, 2021 2 commits
  22. 16 Aug, 2021 2 commits
  23. 30 Jul, 2021 1 commit
    • Deepak Narayanan's avatar
      Support for pipeline parallelism in T5 model · 46c74b4c
      Deepak Narayanan authored
      - Accumulate encoder hidden state gradient to handle skip connection
      - Correctly compute the number of layers in encoder / decoder for T5 model
      - Ensure e weights are initialized the same way in embeddings
      - Synchronize embedding gradients across encoder and decoder for T5 model
      - Support for checkpoint loading and saving
      46c74b4c
  24. 21 Jul, 2021 1 commit
  25. 12 Jul, 2021 1 commit
  26. 05 Jul, 2021 2 commits
  27. 02 Jul, 2021 1 commit
  28. 29 Jun, 2021 1 commit
  29. 03 Apr, 2021 1 commit
  30. 02 Apr, 2021 1 commit
  31. 24 Mar, 2021 1 commit
  32. 19 Mar, 2021 1 commit
  33. 08 Mar, 2021 1 commit
  34. 04 Mar, 2021 1 commit