Code

Auto sft#9728

Code

Open

blacksheep-Aristotle wants to merge 91 commits intoPaddlePaddle:developPaddlePaddle/PaddleNLP:developfrom blacksheep-Aristotle:auto_sftblacksheep-Aristotle/PaddleNLP:auto_sftCopy head branch name to clipboard

+2,689-22

Commits

Commits on Jan 8, 2025

• add single_model network and use intermediate api

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  85d6511

  Copy full SHA for 85d6511

• [AutoParallel]: fix llama_model_network run error

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  e87135d

  Copy full SHA for e87135d

• New version of auto config

  

  FeixLiu

  authored and

  zhangyuqin1998

  committed
  8bb66c9

  Copy full SHA for 8bb66c9

• fix gpt_network to use intermediate_api

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  0f0ad13

  Copy full SHA for 0f0ad13

• fix gpt_network to use intermediate_api

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  802827c

  Copy full SHA for 802827c

• update api

  

  FeixLiu

  authored and

  zhangyuqin1998

  committed
  8427405

  Copy full SHA for 8427405

• update plan

  

  FeixLiu

  authored and

  zhangyuqin1998

  committed
  f150efa

  Copy full SHA for f150efa

• qwen fit base api

  

  FeixLiu

  authored and

  zhangyuqin1998

  committed
  8e9f00d

  Copy full SHA for 8e9f00d

• [AutoParallel]:gpt single network support tp to share_embedding

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  0bed451

  Copy full SHA for 0bed451

• add intermediate ci

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  f18b49d

  Copy full SHA for f18b49d

• add single_model network and use intermediate api

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  18dc01d

  Copy full SHA for 18dc01d

• New version of auto config

  

  FeixLiu

  authored and

  zhangyuqin1998

  committed
  917d995

  Copy full SHA for 917d995

• fix sharding

  

  FeixLiu

  authored and

  zhangyuqin1998

  committed
  573ccd1

  Copy full SHA for 573ccd1

• fix gpt_network to use intermediate_api

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  9fab6f3

  Copy full SHA for 9fab6f3

• fix gpt_network to use intermediate_api

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  1a3c7a0

  Copy full SHA for 1a3c7a0

• update gpt run_pretrain_py

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  64ced7c

  Copy full SHA for 64ced7c

• fix sharding error

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  05b4845

  Copy full SHA for 05b4845

• fix gpt format error

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  dff5db5

  Copy full SHA for dff5db5

• [AutoParallel]:fix llama vpp ci error

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  78d94e5

  Copy full SHA for 78d94e5

• [AutoParallel]:fix ipp error

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  57eb989

  Copy full SHA for 57eb989

• [AutoParallel]:fix a100 ci error

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  a7dfee3

  Copy full SHA for a7dfee3

• [AutoParallel]:fix a100 ci error

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  64cdab4

  Copy full SHA for 64cdab4

• [AutoParallel]:add explanatory note

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  3ec357b

  Copy full SHA for 3ec357b

• Delete =1.0.0

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  c22e514

  Copy full SHA for c22e514

• [AutoParallel]:add run_fintune scripts

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  f71d655

  Copy full SHA for f71d655

• [AutoParallel]:auto parallel support lora model

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  c015b35

  Copy full SHA for c015b35

• update auto_lora_model

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  af1af5c

  Copy full SHA for af1af5c

• update auto_lora_model

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  aa841d4

  Copy full SHA for aa841d4

• [AutoParallel]:nlp support run lora model with intermediate

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  8f8b0a4

  Copy full SHA for 8f8b0a4

• [AutoParallel]:update format

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  ebde991

  Copy full SHA for ebde991

• [AutoParallel]:support input attentionmask

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  3386a3e

  Copy full SHA for 3386a3e

• [AutoParallel]:shard dataloader support multi inputs

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  3a73ad8

  Copy full SHA for 3a73ad8

• [AutoParallel]:auto_sft rebase develop

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  1fd7d7e

  Copy full SHA for 1fd7d7e

• [AutoParallel]:auto_sft rebase develop

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  907f287

  Copy full SHA for 907f287

Commits on Jan 9, 2025

• [AutoParallel]:fix lora parallel mode

  

  blacksheep-Aristotle

  authored and

  zhangyuqin1998

  committed
  fe9e56e

  Copy full SHA for fe9e56e

Commits on Feb 20, 2025

• add single_model network and use intermediate api

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  14a0fc4

  Copy full SHA for 14a0fc4

• [AutoParallel]: fix llama_model_network run error

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  80a676b

  Copy full SHA for 80a676b

• New version of auto config

  

  FeixLiu

  authored and

  liufengwei0103

  committed
  576c5f1

  Copy full SHA for 576c5f1

• fix gpt_network to use intermediate_api

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  6dc3345

  Copy full SHA for 6dc3345

• fix gpt_network to use intermediate_api

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  caa8a50

  Copy full SHA for caa8a50

• update api

  

  FeixLiu

  authored and

  liufengwei0103

  committed
  c3b3ee6

  Copy full SHA for c3b3ee6

• update plan

  

  FeixLiu

  authored and

  liufengwei0103

  committed
  2d38858

  Copy full SHA for 2d38858

• qwen fit base api

  

  FeixLiu

  authored and

  liufengwei0103

  committed
  f869aa0

  Copy full SHA for f869aa0

• [AutoParallel]:gpt single network support tp to share_embedding

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  4c79de3

  Copy full SHA for 4c79de3

• add intermediate ci

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  2b8992b

  Copy full SHA for 2b8992b

• add single_model network and use intermediate api

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  098f454

  Copy full SHA for 098f454

• New version of auto config

  

  FeixLiu

  authored and

  liufengwei0103

  committed
  3ec33ec

  Copy full SHA for 3ec33ec

• fix sharding

  

  FeixLiu

  authored and

  liufengwei0103

  committed
  c867b31

  Copy full SHA for c867b31

• fix gpt_network to use intermediate_api

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  e314a49

  Copy full SHA for e314a49

• fix gpt_network to use intermediate_api

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  4017c25

  Copy full SHA for 4017c25

• update gpt run_pretrain_py

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  63a423d

  Copy full SHA for 63a423d

• fix sharding error

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  882fe73

  Copy full SHA for 882fe73

• fix gpt format error

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  255c1f1

  Copy full SHA for 255c1f1

• [AutoParallel]:fix llama vpp ci error

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  fcb0593

  Copy full SHA for fcb0593

• [AutoParallel]:fix ipp error

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  906c892

  Copy full SHA for 906c892

• [AutoParallel]:fix a100 ci error

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  763c00f

  Copy full SHA for 763c00f

• [AutoParallel]:fix a100 ci error

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  1e9fc7a

  Copy full SHA for 1e9fc7a

• [AutoParallel]:add explanatory note

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  7f5b872

  Copy full SHA for 7f5b872

• Delete =1.0.0

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  ba7a3e7

  Copy full SHA for ba7a3e7

• [AutoParallel]:add run_fintune scripts

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  142d07f

  Copy full SHA for 142d07f

• [AutoParallel]:auto parallel support lora model

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  587a936

  Copy full SHA for 587a936

• update auto_lora_model

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  b108cc0

  Copy full SHA for b108cc0

• update auto_lora_model

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  f535993

  Copy full SHA for f535993

• [AutoParallel]:nlp support run lora model with intermediate

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  24ed043

  Copy full SHA for 24ed043

• [AutoParallel]:update format

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  2e540c6

  Copy full SHA for 2e540c6

• [AutoParallel]:support input attentionmask

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  981c246

  Copy full SHA for 981c246

• [AutoParallel]:shard dataloader support multi inputs

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  bf1c607

  Copy full SHA for bf1c607

• [AutoParallel]:auto_sft rebase develop

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  c259ff3

  Copy full SHA for c259ff3

• [AutoParallel]:auto_sft rebase develop

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  8a0823e

  Copy full SHA for 8a0823e

• [AutoParallel]:fix lora parallel mode

  

  blacksheep-Aristotle

  authored and

  liufengwei0103

  committed
  b8977d4

  Copy full SHA for b8977d4

Commits on Feb 21, 2025

• [AutoParallel]: fix bug about file dep circular and lora config error

  liufengwei0103

  committed
  1d0585b

  Copy full SHA for 1d0585b

• [AutoParallel]: recover default base api config in launch script

  liufengwei0103

  committed
  059ba83

  Copy full SHA for 059ba83

• [AutoParallel]: add to do

  liufengwei0103

  committed
  b3d0fa5

  Copy full SHA for b3d0fa5

• [AutoParallel]: add to do

  liufengwei0103

  committed
  d917f46

  Copy full SHA for d917f46

• Merge branch 'auto_sft' into auto_sft

  liufengwei0103

  authored
  95f7925

  Copy full SHA for 95f7925

• rebase and fix lora config and file dep bug (#28)

  Show description for 211a493
  liufengwei0103

  authored
  211a493

  Copy full SHA for 211a493

Commits on Feb 22, 2025

• [AutoParallel]: add lora ci and fix bug

  liufengwei0103

  committed
  e07595a

  Copy full SHA for e07595a

• [AutoParallel]: add lora ci and fix bug (#29)

  Show description for 965a491
  liufengwei0103

  authored
  965a491

  Copy full SHA for 965a491

• [AutoParallel]: fix lint error

  liufengwei0103

  committed
  6a669f4

  Copy full SHA for 6a669f4

Commits on Feb 24, 2025

• [AutoParallel]: fix lint error

  liufengwei0103

  committed
  57dc1f1

  Copy full SHA for 57dc1f1

• [AutoParallel]: fix lint error

  liufengwei0103

  committed
  5761457

  Copy full SHA for 5761457

• [AutoParallel]: fix ci error about print model

  liufengwei0103

  committed
  79c954e

  Copy full SHA for 79c954e

• [AutoParallel]: fix typo

  liufengwei0103

  committed
  92c0861

  Copy full SHA for 92c0861

• [AutoParallel]: fix

  liufengwei0103

  committed
  cf7bac0

  Copy full SHA for cf7bac0

• [AutoParallel]: fix llama_dygraph_auto_bs8_fp16_DP2-MP2-PP2_intermediate loss base

  liufengwei0103

  committed
  be2cb7b

  Copy full SHA for be2cb7b

• [AutoParallel]: delete typo

  liufengwei0103

  committed
  54a44a1

  Copy full SHA for 54a44a1

• [AutoParallel]: delete comment

  liufengwei0103

  committed
  c77619b

  Copy full SHA for c77619b

• [AutoParallel]: fix ci error not on a100

  liufengwei0103

  committed
  4413a72

  Copy full SHA for 4413a72

Commits on Feb 25, 2025

• [AutoParallel]: delete comment

  liufengwei0103

  committed
  f0352ad

  Copy full SHA for f0352ad

• [AutoParallel]: delete useless code

  liufengwei0103

  committed
  641bcc7

  Copy full SHA for 641bcc7

• [AutoParallel]: ddelete launch script

  liufengwei0103

  committed
  7309602

  Copy full SHA for 7309602