Commits

Branch selector

main

User selector

All users

All time

Commit history

Commits on Jan 2, 2024

• add: support for `peft` in ddpo. (#1165)

  Show description for 20428c4
  sayakpaul

  authored
  20428c4View commit details

  Copy full SHA for 20428c4

  Browse repository at this point

Commits on Dec 29, 2023

• Minor fixes to some comments in some examples. (#1156)

  mattholl

  authored
  6614b8aView commit details

  Copy full SHA for 6614b8a

  Browse repository at this point

• change device order of metrics (#1154)

  brcps12

  authored
  df7b770View commit details

  Copy full SHA for df7b770

  Browse repository at this point

Commits on Dec 27, 2023

• SFT Tokenizer Fix (#1142)

  ChrisCates

  authored
  18a33ffView commit details

  Copy full SHA for 18a33ff

  Browse repository at this point

Commits on Dec 26, 2023

• [`xxxTrainer`] Add unsloth tag (#1130)

  Show description for 911d365
  younesbelkada

  authored
  911d365View commit details

  Copy full SHA for 911d365

  Browse repository at this point

• add peft_module_casting_to_bf16 in DPOTrainer (#1143)

  Show description for 95ec857
  

  sywangyi

  and

  kashif

  authored
  95ec857View commit details

  Copy full SHA for 95ec857

  Browse repository at this point

• set dev version (#1145)

  younesbelkada

  authored
  3539f3eView commit details

  Copy full SHA for 3539f3e

  Browse repository at this point

• Release: v0.7.7 (#1144)

  younesbelkada

  authored
  e451298View commit details

  Copy full SHA for e451298

  Browse repository at this point

• [`PPOTrainer` / `DDPOTrainer`] Fix ppo & ddpo push to Hub (#1141)

  Show description for 3efb484
  younesbelkada

  authored
  3efb484View commit details

  Copy full SHA for 3efb484

  Browse repository at this point

Commits on Dec 23, 2023

• reformatted (#1128)

  zhangsibo1129

  authored
  8f5b492View commit details

  Copy full SHA for 8f5b492

  Browse repository at this point

• reformatted (#1129)

  zhangsibo1129

  authored
  e0dec27View commit details

  Copy full SHA for e0dec27

  Browse repository at this point

Commits on Dec 22, 2023

• Add type hints to core.py (#1097)

  Show description for 6ef785a
  zachschillaci27

  authored
  6ef785aView commit details

  Copy full SHA for 6ef785a

  Browse repository at this point

• clear up the parameters of supervised_finetuning.py (#1126)

  Show description for 950ee21
  sywangyi

  authored
  950ee21View commit details

  Copy full SHA for 950ee21

  Browse repository at this point

• set dev version (#1135)

  younesbelkada

  authored
  c1bb1f3View commit details

  Copy full SHA for c1bb1f3

  Browse repository at this point

• Release: v0.7.6 (#1134)

  younesbelkada

  authored
  54babd9View commit details

  Copy full SHA for 54babd9

  Browse repository at this point

• [`xxxTrainer`] multi-tags support for tagging (#1133)

  Show description for 0c4edb7
  younesbelkada

  authored
  0c4edb7View commit details

  Copy full SHA for 0c4edb7

  Browse repository at this point

• set dev version (#1132)

  younesbelkada

  authored
  17ec68dView commit details

  Copy full SHA for 17ec68d

  Browse repository at this point

• Release: v0.7.5 (#1131)

  younesbelkada

  authored
  9be5680View commit details

  Copy full SHA for 9be5680

  Browse repository at this point

• [`Docs`] Add unsloth optimizations in TRL's documentation (#1119)

  Show description for f11e213
  

  younesbelkada

  and

  danielhanchen

  authored
  f11e213View commit details

  Copy full SHA for f11e213

  Browse repository at this point

• rename kto loss (#1127)

  kashif

  authored
  814fe39View commit details

  Copy full SHA for 814fe39

  Browse repository at this point

Commits on Dec 21, 2023

• save eval_dataset for subsequent calls (#1125)

  kashif

  authored
  06b7959View commit details

  Copy full SHA for 06b7959

  Browse repository at this point

• [`xxxTrainer`] Add tags to all trainers in TRL (#1120)

  Show description for b07935f
  younesbelkada

  authored
  b07935fView commit details

  Copy full SHA for b07935f

  Browse repository at this point

• Update `description` in `setup.py` (#1101)

  alvarobartt

  authored
  2aff709View commit details

  Copy full SHA for 2aff709

  Browse repository at this point

Commits on Dec 20, 2023

• fix gradient checkpointing when using PEFT (#1118)

  pacman100

  authored
  830cadfView commit details

  Copy full SHA for 830cadf

  Browse repository at this point

• Make prepending of bos token configurable. (#1114)

  Show description for f2acd82
  

  pacman100

  and

  younesbelkada

  authored
  f2acd82View commit details

  Copy full SHA for f2acd82

  Browse repository at this point

Commits on Dec 19, 2023

• `peft_module_casting_to_bf16` util method, `append_concat_token` flag, remove callback `PeftSavingCallback` (#1110)

  Show description for f100ca3
  pacman100

  authored
  f100ca3View commit details

  Copy full SHA for f100ca3

  Browse repository at this point

Commits on Dec 15, 2023

• [Feature] Add Ascend NPU accelerator support (#1096)

  Show description for d708ec2
  ji-huazhong

  authored
  d708ec2View commit details

  Copy full SHA for d708ec2

  Browse repository at this point

• Updated documentation for docs/source/reward_trainer.mdx to import the correct Enum for the reward modelling LoRA config (#1092)

  cm2435

  authored
  8140129View commit details

  Copy full SHA for 8140129

  Browse repository at this point

Commits on Dec 12, 2023

• [DPO] use ref model logprobs if it exists in the data (#885)

  Show description for 48b3ef0
  

  4 peopleauthored
  48b3ef0View commit details

  Copy full SHA for 48b3ef0

  Browse repository at this point

• consistency on log (#1084)

  tcapelle

  authored
  c0ce52aView commit details

  Copy full SHA for c0ce52a

  Browse repository at this point

Commits on Dec 11, 2023

• Removing `tyro` in `sft_llama2.py` (#1081)

  Show description for 393dbf6
  vwxyzjn

  authored
  393dbf6View commit details

  Copy full SHA for 393dbf6

  Browse repository at this point

• Make CI happy (#1080)

  Show description for 94fa4b0
  younesbelkada

  authored
  94fa4b0View commit details

  Copy full SHA for 94fa4b0

  Browse repository at this point

• add local folder support as input for rl_training. (#1078)

  Show description for cb7819e
  sywangyi

  authored
  cb7819eView commit details

  Copy full SHA for cb7819e

  Browse repository at this point

• Add args to SFT example (#1079)

  lewtun

  authored
  8f0fc4cView commit details

  Copy full SHA for 8f0fc4c

  Browse repository at this point

• [DPO] add KTO loss (#1075)

  Show description for d275cb4
  

  kashif

  and

  lewtun

  authored
  d275cb4View commit details

  Copy full SHA for d275cb4

  Browse repository at this point