Code

[Auto Parallel]Update AutoTuner#56939

Code

Merged

zhaoyinglia merged 17 commits intoPaddlePaddle:developPaddlePaddle/Paddle:developfrom Caozhou1995:add_vpp_searchCaozhou1995/Paddle:add_vpp_searchCopy head branch name to clipboardSep 25, 2023

+598-171

Commits

Commits on Sep 12, 2023

• add vpp search

  Caozhou1995

  committed
  bcc7832

  Copy full SHA for bcc7832

• support cmd json or yaml

  Caozhou1995

  committed
  77e7468

  Copy full SHA for 77e7468

• add auto tuner log

  Caozhou1995

  committed
  5ba1bda

  Copy full SHA for 5ba1bda

• add OOM prune

  Caozhou1995

  committed
  94b4121

  Copy full SHA for 94b4121

• support sft and lora best cfg

  Caozhou1995

  committed
  e11e74a

  Copy full SHA for e11e74a

• support sp

  Caozhou1995

  committed
  e8b958b

  Copy full SHA for e8b958b

• fix sft/lora cfg

  Caozhou1995

  committed
  712be57

  Copy full SHA for 712be57

• fix json/yaml bug and update multi nodes status

  Caozhou1995

  committed
  36523fe

  Copy full SHA for 36523fe

• add pp and mp prune flag

  Caozhou1995

  committed
  608afea

  Copy full SHA for 608afea

• update log and csv path

  Caozhou1995

  committed
  9407f96

  Copy full SHA for 9407f96

• merge dev

  Caozhou1995

  committed
  4fb4af0

  Copy full SHA for 4fb4af0

Commits on Sep 14, 2023

• speed prune

  Caozhou1995

  committed
  a79e271

  Copy full SHA for a79e271

Commits on Sep 15, 2023

• add search stage and run best stage cmd

  Caozhou1995

  committed
  16639ce

  Copy full SHA for 16639ce

• fix recompute_granularity prune bug

  Caozhou1995

  committed
  fda5145

  Copy full SHA for fda5145

Commits on Sep 18, 2023

• fix get best cfg in sft and lora mode bug

  Caozhou1995

  committed
  fe82db4

  Copy full SHA for fe82db4

Commits on Sep 21, 2023

• fix single card bug

  Caozhou1995

  committed
  9a06c25

  Copy full SHA for 9a06c25

Commits on Sep 25, 2023

• update read metric

  Caozhou1995

  committed
  e3a0038

  Copy full SHA for e3a0038