Skip to content

GitLab

Switch branch/tag
  1. 21 Dec, 2021 1 commit
  3. 06 Dec, 2021 1 commit
  5. 03 Dec, 2021 1 commit
  7. 20 Nov, 2021 1 commit
  9. 19 Nov, 2021 1 commit
  11. 17 Nov, 2021 1 commit
  13. 16 Nov, 2021 1 commit
  15. 11 Nov, 2021 1 commit
  17. 04 Nov, 2021 1 commit
  19. 30 Sep, 2021 1 commit
  21. 30 Aug, 2021 1 commit
  23. 27 Aug, 2021 1 commit
  25. 19 Aug, 2021 1 commit
  27. 06 Aug, 2021 1 commit
  29. 28 Jul, 2021 1 commit
  31. 20 Jul, 2021 1 commit
  33. 24 Jun, 2021 1 commit
  35. 03 Jun, 2021 1 commit
  37. 02 Jun, 2021 1 commit
  39. 07 May, 2021 1 commit
  41. 13 Apr, 2021 1 commit
    • Reed Wanderman-Milne's avatar
      Use nonexperimental mixed precision API. · c0ac8d1c
      Reed Wanderman-Milne authored 
      This replaces symbols in tf.keras.mixed_precision.experimental with the corresponding nonexperimental symbols. In some cases, passing a Policy is replaced with passing a policy name for conciseness.

Additionally, for the Shakespeare model, the loss_scale flag is removed, since supporting it with the nonexperimental API is slightly more verbose and it is recommended users use the default loss scale.

PiperOrigin-RevId: 368123944
      c0ac8d1c
  43. 12 Apr, 2021 1 commit
    • Reed Wanderman-Milne's avatar
      Use nonexperimental mixed precision API for official models. · e6cda015
      Reed Wanderman-Milne authored 
      For all modified calls to set_mixed_precision_policy(), the loss_scale argument was removed, as it cannot be passed if the nonexperimental API is used. For all such callers, the loss_scale is later used to explicitly create a LossScaleOptimizer, so removing the argument has no impact.

Switching to the non-experimental LossScaleOptimizer has no effect, as it has near identical behavior and all isinstance checks within the official models check for the non-experimental version.

PiperOrigin-RevId: 368101975
      e6cda015
  45. 09 Apr, 2021 1 commit
    • Reed Wanderman-Milne's avatar
      Remove dynamic_loss_scale argument to define_performance. · 20ed8cf0
      Reed Wanderman-Milne authored 
      All models which support loss scaling support dynamic loss scaling, so the argument has no purpose. It used to be that some models scaled the loss manually instead of using a LossScaleOptimizer, and so did not support dynamic loss scaling.

PiperOrigin-RevId: 367719521
      20ed8cf0
  47. 06 Apr, 2021 1 commit
  49. 25 Mar, 2021 1 commit
  51. 23 Mar, 2021 1 commit
  53. 20 Mar, 2021 1 commit
  55. 16 Mar, 2021 1 commit
  57. 04 Mar, 2021 1 commit
  59. 03 Mar, 2021 1 commit
  61. 02 Mar, 2021 1 commit
  63. 01 Mar, 2021 2 commits
  65. 03 Feb, 2021 1 commit
  67. 17 Nov, 2020 1 commit
  69. 10 Nov, 2020 1 commit
  71. 12 Oct, 2020 2 commits
  73. 28 Sep, 2020 1 commit
  75. 22 Sep, 2020 1 commit