Code

Add checkpoint unit test#35

Code

Merged

sfc-gh-mwyatt merged 9 commits intomainsnowflakedb/ArcticTraining:mainfrom mwyatt/add-ckpt-testsnowflakedb/ArcticTraining:mwyatt/add-ckpt-testCopy head branch name to clipboardJan 30, 2025

+248-42

Commits

Commits on Jan 22, 2025

• add initial checkpoint test

  sfc-gh-mwyatt

  committed
  d67b4ea

  Copy full SHA for d67b4ea

Commits on Jan 23, 2025

• avoid using subprocess.run in test calls as it can hide error messages

  sfc-gh-mwyatt

  committed
  ce2c1af

  Copy full SHA for ce2c1af

Commits on Jan 24, 2025

• cpu / gpu dist env setup

  sfc-gh-mwyatt

  committed
  4fbc4e7

  Copy full SHA for 4fbc4e7

• add ds engine test

  sfc-gh-mwyatt

  committed
  d260875

  Copy full SHA for d260875

• cleanup

  sfc-gh-mwyatt

  committed
  060056b

  Copy full SHA for 060056b

• Merge branch 'main' into mwyatt/add-ckpt-test

  sfc-gh-mwyatt

  authored
  e289d16

  Copy full SHA for e289d16

• add process group destruction

  sfc-gh-mwyatt

  committed
  63c109a

  Copy full SHA for 63c109a

• fix error with setting cuda state when loading checkpoint

  sfc-gh-mwyatt

  committed
  23e32ea

  Copy full SHA for 23e32ea

• fix bug when saving checkpoint on non-CUDA torch

  sfc-gh-mwyatt

  committed
  29ba3b2

  Copy full SHA for 29ba3b2