Commits

Branch selector

 master

User selector

definitelynotmcarilli

Datepicker

All time

Commit History

Commits on Dec 3, 2019

• Don't check if distributed is initialized on Windows

  definitelynotmcarilli

  committed
  f37fdf0

  Copy full SHA for f37fdf0

Commits on Oct 22, 2019

• Making the encouragement to use O1 a bit stronger...

  definitelynotmcarilli

  committed
  95d6c00

  Copy full SHA for 95d6c00

Commits on Oct 10, 2019

• Adding presentation link to sphinx landing page

  definitelynotmcarilli

  committed
  0889859

  Copy full SHA for 0889859

Commits on Aug 27, 2019

• Deleting test_fp16_optimizer.py

  definitelynotmcarilli

  committed
  30ed793

  Copy full SHA for 30ed793

• Docstring updates

  definitelynotmcarilli

  committed
  b9e5d37

  Copy full SHA for b9e5d37

• Docstring updates

  definitelynotmcarilli

  committed
  17e8a55

  Copy full SHA for 17e8a55

• Merge branch 'master' of https://github.com/NVIDIA/apex

  definitelynotmcarilli

  committed
  ea7c209

  Copy full SHA for ea7c209

• Updating docstrings for fused optimizers

  definitelynotmcarilli

  committed
  427e82c

  Copy full SHA for 427e82c

• Merge branch 'FDecaYed-deyuf/fused_optimizer_v2'

  definitelynotmcarilli

  committed
  1564802

  Copy full SHA for 1564802

Commits on Jul 3, 2019

• Pulling in deprecation warning changes

  definitelynotmcarilli

  committed
  665b2dd

  Copy full SHA for 665b2dd

• Bumping container version for test

  definitelynotmcarilli

  committed
  d352d44

  Copy full SHA for d352d44

• Remove deprecated Type.h

  definitelynotmcarilli

  committed
  816813f

  Copy full SHA for 816813f

• Bumping container version for test

  definitelynotmcarilli

  committed
  1483f22

  Copy full SHA for 1483f22

• Remove deprecated Type.h

  definitelynotmcarilli

  committed
  7096b1b

  Copy full SHA for 7096b1b

• Changing AT_CHECK to TORCH_CHECK

  definitelynotmcarilli

  committed
  adee29f

  Copy full SHA for adee29f

• Fix use of multi_tensor_l2norm, remove test using deprecated syntax

  definitelynotmcarilli

  committed
  b9336b1

  Copy full SHA for b9336b1

Commits on Jul 2, 2019

• cosmetic

  definitelynotmcarilli

  committed
  47da14a

  Copy full SHA for 47da14a

• Merging in master

  definitelynotmcarilli

  committed
  8a32e42

  Copy full SHA for 8a32e42

Commits on Jun 27, 2019

• Clarifying documentation on gradient accumulation

  definitelynotmcarilli

  committed
  18f2eae

  Copy full SHA for 18f2eae

Commits on Jun 25, 2019

• grid_sample should be fp32 for now

  definitelynotmcarilli

  committed
  656d14b

  Copy full SHA for 656d14b

Commits on Jun 24, 2019

• Merge branch 'master' of https://github.com/NVIDIA/apex

  definitelynotmcarilli

  committed
  9ce8017

  Copy full SHA for 9ce8017

• Docstring for multiple losses

  definitelynotmcarilli

  committed
  f855756

  Copy full SHA for f855756

• Updating gradient accumulation guidance

  definitelynotmcarilli

  committed
  ca35aa7

  Copy full SHA for ca35aa7

Commits on Jun 21, 2019

• Make main_amp.py more profiling-friendly

  definitelynotmcarilli

  committed
  f29b3f8

  Copy full SHA for f29b3f8

• Don't need to blacklist mean for pytorch >= 1.1

  definitelynotmcarilli

  committed
  4b9858e

  Copy full SHA for 4b9858e

Commits on Jun 20, 2019

• Fix end-of-epoch with record_stream

  definitelynotmcarilli

  committed
  90e5b05

  Copy full SHA for 90e5b05

Commits on Jun 19, 2019

• Merge branch 'master' of https://github.com/NVIDIA/apex

  definitelynotmcarilli

  committed
  1ccaaf4

  Copy full SHA for 1ccaaf4

• Fix for NVIDIA#361

  definitelynotmcarilli

  committed
  68c850d

  Copy full SHA for 68c850d

Commits on Jun 18, 2019

• Fix rare caching allocator race condition in imagenet prefetcher

  definitelynotmcarilli

  committed
  d5e2bb4

  Copy full SHA for d5e2bb4

Commits on Jun 17, 2019

• More helpful message for unexpected opt_level

  definitelynotmcarilli

  committed
  c3bcf18

  Copy full SHA for c3bcf18

Commits on Jun 14, 2019

• Merge branch 'master' of https://github.com/NVIDIA/apex

  definitelynotmcarilli

  committed
  a9f5f71

  Copy full SHA for a9f5f71

• Removing gradient_average_split_factor

  definitelynotmcarilli

  committed
  41c9851

  Copy full SHA for 41c9851

• Adding delay_overflow_check=False ninja control point

  definitelynotmcarilli

  committed
  4a9c2a5

  Copy full SHA for 4a9c2a5

Commits on Jun 13, 2019

• disable_allreduce -> _disable_allreduce

  definitelynotmcarilli

  committed
  ae7f0de

  Copy full SHA for ae7f0de

Commits on Jun 11, 2019

• Allow multi_tensor_lamb to update fp16 params

  definitelynotmcarilli

  committed
  47e3367

  Copy full SHA for 47e3367

Pagination

Previous

Next