Commits

Breadcrumbs

History for

onmaster

User selector

All users

Datepicker

All time

Commit History

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 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 Sep 6, 2019

• Fix for NVIDIA#456 (NVIDIA#477)

  Show description for 325f5a0
  mcarilli

  authored
  325f5a0

  Copy full SHA for 325f5a0

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

Commits on Aug 27, 2019

• Enable Checkpointing (NVIDIA#420)

  Show description for dec4fdd
  

  ptrblck

  authored and

  mcarilli

  committed
  dec4fdd

  Copy full SHA for dec4fdd

Commits on Aug 17, 2019

• disable breaking test until switch to test against upstream v1.2.0

  FDecaYed

  committed
  f855f85

  Copy full SHA for f855f85

Commits on Aug 13, 2019

• Merge branch 'master' into deyuf/fused_optimizer_v2

  FDecaYed

  committed
  aee5aff

  Copy full SHA for aee5aff

• Reverse to Fused* naming, clean up accordingly:

  Show description for 007c594
  FDecaYed

  committed
  007c594

  Copy full SHA for 007c594

• Adding PyProf to Apex (NVIDIA#404)

  Show description for 880ab92
  

  2 peopleauthored and

  mcarilli

  committed
  880ab92

  Copy full SHA for 880ab92

Commits on Aug 12, 2019

• keep old fused* name and rename new optimizers without prefix

  FDecaYed

  committed
  adad599

  Copy full SHA for adad599

• Merge branch 'multi_tensor_sgd' into deyuf/fused_optimizer_v2

  FDecaYed

  committed
  4d6ed50

  Copy full SHA for 4d6ed50

Commits on Aug 8, 2019

• initial commit to make fused optimizers compatible with AMP

  FDecaYed

  committed
  690b1f7

  Copy full SHA for 690b1f7

Commits on Aug 6, 2019

• Clean up layer norm tests (NVIDIA#418)

  Show description for 3ef01fa
  

  ngimel

  authored and

  mcarilli

  committed
  3ef01fa

  Copy full SHA for 3ef01fa

Commits on Jul 26, 2019

• [sbn update] (NVIDIA#384)

  Show description for 896ecdd
  

  jjsjann123

  authored and

  syed-ahmed

  committed
  896ecdd

  Copy full SHA for 896ecdd

Commits on Jul 12, 2019

• [sbn update] (NVIDIA#384)

  Show description for 574fe24
  

  jjsjann123

  authored and

  mcarilli

  committed
  574fe24

  Copy full SHA for 574fe24

Commits on Jul 3, 2019

• Bumping container version for test

  definitelynotmcarilli

  committed
  d352d44

  Copy full SHA for d352d44

• Bumping container version for test

  definitelynotmcarilli

  committed
  1483f22

  Copy full SHA for 1483f22

• Fix use of multi_tensor_l2norm, remove test using deprecated syntax

  definitelynotmcarilli

  committed
  b9336b1

  Copy full SHA for b9336b1

Commits on Jul 2, 2019

• Merging in master

  definitelynotmcarilli

  committed
  8a32e42

  Copy full SHA for 8a32e42

Commits on May 31, 2019

• Give multi-tensor L2 norm the ability to compute norms per-tensor as well as globally (NVIDIA#333)

  Show description for 93338e6
  mcarilli

  authored
  93338e6

  Copy full SHA for 93338e6

Commits on May 27, 2019

• test cleanup

  definitelynotmcarilli

  committed
  d68ec71

  Copy full SHA for d68ec71

• FusedSGD tests passing for all opt_levels

  definitelynotmcarilli

  committed
  848c777

  Copy full SHA for 848c777

Commits on May 23, 2019

• Merging in latest master changes

  definitelynotmcarilli

  committed
  c142714

  Copy full SHA for c142714

Commits on May 16, 2019

• Support add_param_group (NVIDIA#310)

  Show description for 4d325d2
  mcarilli

  authored
  4d325d2

  Copy full SHA for 4d325d2

Commits on May 13, 2019

• Adding docker build test for 1.1 container

  definitelynotmcarilli

  committed
  f2b3a62

  Copy full SHA for f2b3a62

Commits on May 2, 2019

• Adding test_fused_sgd.py

  definitelynotmcarilli

  committed
  d050543

  Copy full SHA for d050543

Commits on May 1, 2019

• allreduce_different_streams is now hidden

  definitelynotmcarilli

  committed
  72bce16

  Copy full SHA for 72bce16

Commits on Apr 30, 2019

• Remove unused tensor in fast_collate (NVIDIA#281)

  Show description for ca2baff
  

  ptrblck

  authored and

  mcarilli

  committed
  ca2baff

  Copy full SHA for ca2baff

Commits on Apr 11, 2019

• Merge branch 'master' into prepare_fused

  definitelynotmcarilli

  committed
  3c53cf8

  Copy full SHA for 3c53cf8

• Fixing merge conflict in setup.py

  definitelynotmcarilli

  committed
  dda5935

  Copy full SHA for dda5935

Commits on Apr 10, 2019

• add new tests to run_test.py

  LamDang

  committed
  6d40465

  Copy full SHA for 6d40465

• quick fix: make FusedLayerNorm compatible with cpu

  LamDang

  committed
  d130ec1

  Copy full SHA for d130ec1

• Kernel + sizes stress test

  definitelynotmcarilli

  committed
  1a48b26

  Copy full SHA for 1a48b26

Commits on Apr 4, 2019

• WIP: Handle arbitrary combinations of optimizers/models/losses (NVIDIA#232)

  Show description for 3f87614
  mcarilli

  authored
  3f87614

  Copy full SHA for 3f87614

Commits on Mar 22, 2019

• Check cuda version (NVIDIA#216)

  Show description for 5b8faa2
  mcarilli

  authored
  5b8faa2

  Copy full SHA for 5b8faa2

Pagination

Previous

Next