Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Dec 18, 2019

• fix beta2 => beta1 (NVIDIA#661)

  

  bonlime

  authored and

  mcarilli

  committed
  0ce8ad3

  Copy full SHA for 0ce8ad3

• updated apex.contrib.optimizers.FP16_Optimizer and FusedSGD (NVIDIA#657)

  

  kexinyu

  authored and

  mcarilli

  committed
  c19ee27

  Copy full SHA for c19ee27

Commits on Dec 5, 2019

• Fixing typo in PyProf README (NVIDIA#637)

  

  ntenenz

  authored and

  mcarilli

  committed
  4ad9b3b

  Copy full SHA for 4ad9b3b

Commits on Dec 3, 2019

• Don't check if distributed is initialized on Windows

  definitelynotmcarilli

  committed
  f37fdf0

  Copy full SHA for f37fdf0

Commits on Nov 22, 2019

• update _amp_state to check distributed on maybe_print (NVIDIA#620)

  

  rmrao

  authored and

  mcarilli

  committed
  82dac9c

  Copy full SHA for 82dac9c

Commits on Nov 6, 2019

• fixing batchnorm 1d input (NVIDIA#590)

  

  jjsjann123

  authored and

  mcarilli

  committed
  37cdaf4

  Copy full SHA for 37cdaf4

Commits on Oct 30, 2019

• Update README.md

  mcarilli

  authored
  606c3dc

  Copy full SHA for 606c3dc

Commits on Oct 23, 2019

• add gelu activation to fp32 list (NVIDIA#564)

  

  Bram Vanroy

  authored and

  mcarilli

  committed
  5b29cc1

  Copy full SHA for 5b29cc1

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 19, 2019

• Made the patched optimizer step function a full method, not simply a function stored as an instance member (NVIDIA#553)

  

  aronhoff

  authored and

  mcarilli

  committed
  4b91326

  Copy full SHA for 4b91326

Commits on Oct 10, 2019

• Adding presentation link to sphinx landing page

  definitelynotmcarilli

  committed
  0889859

  Copy full SHA for 0889859

• Adding links to references. (maybe make this a subrepo?)

  mcarilli

  authored
  ec93c75

  Copy full SHA for ec93c75

Commits on Oct 9, 2019

• allow for non-distributed envs (Windows) (NVIDIA#531)

  

  Bram Vanroy

  authored and

  mcarilli

  committed
  fab319f

  Copy full SHA for fab319f

• Fixed tensor core lookup for Turing (NVIDIA#534)

  

  mkolod

  authored and

  mcarilli

  committed
  753c427

  Copy full SHA for 753c427

Commits on Oct 8, 2019

• Include loss scaling in README code example (NVIDIA#523)

  

  f0k

  authored and

  mcarilli

  committed
  e87b579

  Copy full SHA for e87b579

Commits on Oct 4, 2019

• move previous fused_adam and fp16_optimizer to contrib (NVIDIA#517)

  Show description for 1904e48
  

  FDecaYed

  authored and

  mcarilli

  committed
  1904e48

  Copy full SHA for 1904e48

Commits on Oct 3, 2019

• Disable tests for mixed opt_levels, add bitwise accurate test of parameters (NVIDIA#520)

  Show description for 0b74bfd
  

  ptrblck

  authored and

  mcarilli

  committed
  0b74bfd

  Copy full SHA for 0b74bfd

Commits on Oct 2, 2019

• fix facebookresearch/maskrcnn-benchmark#802 (NVIDIA#516)

  

  timotheecour

  authored and

  mcarilli

  committed
  03421e8

  Copy full SHA for 03421e8

Commits on Sep 13, 2019

• Seems to work locally (NVIDIA#490)

  mcarilli

  authored
  3ae89c7

  Copy full SHA for 3ae89c7

Commits on Sep 12, 2019

• Fixed error in convert_syncbn_model function (NVIDIA#380)

  

  whyjay

  authored and

  mcarilli

  committed
  e6cb749

  Copy full SHA for e6cb749

Commits on Sep 11, 2019

• removing nvtx range used for debugging (NVIDIA#485)

  

  jjsjann123

  authored and

  mcarilli

  committed
  ad98cc5

  Copy full SHA for ad98cc5

Commits on Sep 6, 2019

• Fix for NVIDIA#456 (NVIDIA#477)

  Show description for 325f5a0
  mcarilli

  authored
  325f5a0

  Copy full SHA for 325f5a0

• LARC needs no Variable (NVIDIA#461)

  Show description for 1bf0d8d
  

  Tony-Y

  authored and

  mcarilli

  committed
  1bf0d8d

  Copy full SHA for 1bf0d8d

Commits on Sep 3, 2019

• Fix issues in fused_dam (NVIDIA#469)

  Show description for 7fa7492
  

  FDecaYed

  authored and

  mcarilli

  committed
  7fa7492

  Copy full SHA for 7fa7492

• remove deprecated backaned.FunctionBackend calls (NVIDIA#466)

  

  ptrblck

  authored and

  mcarilli

  committed
  35a8578

  Copy full SHA for 35a8578

Commits on Aug 30, 2019

• [novograd] move exp_avg_sq to param device in load_state_dict (NVIDIA#459)

  

  FDecaYed

  authored and

  mcarilli

  committed
  53eae19

  Copy full SHA for 53eae19

Commits on Aug 27, 2019

• Enable Checkpointing (NVIDIA#420)

  Show description for dec4fdd
  

  ptrblck

  authored and

  mcarilli

  committed
  dec4fdd

  Copy full SHA for dec4fdd

• 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 Aug 25, 2019

• skip instancenorm in convert_syncbn_model (NVIDIA#438)

  

  diggerdu

  authored and

  mcarilli

  committed
  78c38db

  Copy full SHA for 78c38db

Commits on Aug 20, 2019

• add back lamb stage1/2 to amp_C python

  FDecaYed

  committed
  b9f0995

  Copy full SHA for b9f0995

Pagination

Previous

Next