Code

some fixes to make training work#5

Code

Open

sfc-gh-rsamdani wants to merge 13 commits intooptim-linearSnowflake-Labs/DeepSpeed:optim-linearfrom rajhans/fixesSnowflake-Labs/DeepSpeed:rajhans/fixesCopy head branch name to clipboard

+3-3

Commits

Commits on Apr 2, 2024

• ensure capacity does not exceed number of tokens

  jeffra

  committed
  d6d1bfe

• auto convert moe param groups

  jeffra

  committed
  c96e350

• remove unrelated test

  jeffra

  committed
  e942e76

• Update test_moe.py

  jeffra

  authored
  2c42ef1

• Merge branch 'master' into auto-param-group

  jeffra

  authored
  5905f9b

Commits on Apr 4, 2024

• Merge pull request #3 from Snowflake-Labs/optim-linear

  Show description for a876db7
  sfc-gh-jrasley

  authored
  a876db7

• Merge branch 'master' into auto-param-group

  jeffra

  authored
  955e37c

• Merge pull request #4 from Snowflake-Labs/auto-param-group

  Show description for fb67b34
  sfc-gh-jrasley

  authored
  fb67b34

Commits on Apr 17, 2024

• some fixes to make lora training work

  sfc-gh-rsamdani

  committed
  8a34783

• small

  sfc-gh-rsamdani

  committed
  5adb928

Commits on Apr 21, 2024

• small

  sfc-gh-rsamdani

  committed
  f8fc11f

• small

  sfc-gh-rsamdani

  committed
  ebcdead

• small

  sfc-gh-rsamdani

  committed
  35af944