Commits

Branch selector

main

User selector

All users

All time

Commit history

Commits on Jun 11, 2024

• added docs and tests

  

  Igor Melnyk

  committed
  83b6c34View commit details

  Copy full SHA for 83b6c34

  Browse repository at this point

Commits on Jun 6, 2024

• Applied format changes

  

  Igor Melnyk

  committed
  1724d82View commit details

  Copy full SHA for 1724d82

  Browse repository at this point

• Merge remote-tracking branch 'upstream/main'

  

  Igor Melnyk

  committed
  732e9cfView commit details

  Copy full SHA for 732e9cf

  Browse repository at this point

• 0.9.4 release (#1708)

  vwxyzjn

  authored
  974b0d3View commit details

  Copy full SHA for 974b0d3

  Browse repository at this point

• SFTTrainer: Fix backward Compatibility issue with `TrainingArguments` (#1707)

  Show description for 39a7d1c
  younesbelkada

  authored
  39a7d1cView commit details

  Copy full SHA for 39a7d1c

  Browse repository at this point

• Fixed doc string and docs for the SFTConfig update (#1706)

  GuilhermeFreire

  authored
  0bdc638View commit details

  Copy full SHA for 0bdc638

  Browse repository at this point

Commits on Jun 5, 2024

• adds AOT

  

  Igor Melnyk

  committed
  64203ddView commit details

  Copy full SHA for 64203dd

  Browse repository at this point

• 0.9.3 release (#1699)

  vwxyzjn

  authored
  275d33bView commit details

  Copy full SHA for 275d33b

  Browse repository at this point

• Update sft_trainer.py (#1698)

  younesbelkada

  authored
  c0819eeView commit details

  Copy full SHA for c0819ee

  Browse repository at this point

• Release 0.9.2 (#1697)

  Show description for a03e7cc
  vwxyzjn

  authored
  a03e7ccView commit details

  Copy full SHA for a03e7cc

  Browse repository at this point

• Quick fix on GPT4-eval (#1696)

  Show description for a13cb89
  vwxyzjn

  authored
  a13cb89View commit details

  Copy full SHA for a13cb89

  Browse repository at this point

Commits on Jun 3, 2024

• Fix typo in DPOTrainer's warnings (#1688)

  qgallouedec

  authored
  84156f1View commit details

  Copy full SHA for 84156f1

  Browse repository at this point

• Skip packing validation (#1673)

  Show description for 4eb0b90
  alex-jw-brooks

  authored
  4eb0b90View commit details

  Copy full SHA for 4eb0b90

  Browse repository at this point

• Fix overriding optimize_device_cache with optimize_cuda_cache in PPOConfig (#1690)

  Show description for 6c203f9
  alexisrozhkov

  authored
  6c203f9View commit details

  Copy full SHA for 6c203f9

  Browse repository at this point

• intial RPO loss (#1686)

  Show description for f18253b
  kashif

  authored
  f18253bView commit details

  Copy full SHA for f18253b

  Browse repository at this point

Commits on May 29, 2024

• Fix max completion length (#1588)

  samuki

  authored
  151a452View commit details

  Copy full SHA for 151a452

  Browse repository at this point

• fix (#1678)

  younesbelkada

  authored
  488b502View commit details

  Copy full SHA for 488b502

  Browse repository at this point

Commits on May 27, 2024

• fix dataset load error (#1670)

  Show description for 3c0a10b
  sywangyi

  authored
  3c0a10bView commit details

  Copy full SHA for 3c0a10b

  Browse repository at this point

Commits on May 24, 2024

• FIX / PPO: Fix `enable_input_require_grads` issues with PPO models (#1664)

  Show description for b031adf
  younesbelkada

  authored
  b031adfView commit details

  Copy full SHA for b031adf

  Browse repository at this point

Commits on May 23, 2024

• Fix ppov2 test case (#1661)

  Show description for e7cb597
  vwxyzjn

  authored
  e7cb597View commit details

  Copy full SHA for e7cb597

  Browse repository at this point

• update eval_strategy (#1662)

  kashif

  authored
  bc8dfbfView commit details

  Copy full SHA for bc8dfbf

  Browse repository at this point

• do not upcast adapters when using FSDP+QLoRA (#1654)

  pacman100

  authored
  e4ed7a3View commit details

  Copy full SHA for e4ed7a3

  Browse repository at this point

• 🤫 TR-DPO implementation (#1593)

  Show description for 9a7efbd
  

  4 peopleauthored
  9a7efbdView commit details

  Copy full SHA for 9a7efbd

  Browse repository at this point

• [DPO] Add 'robust' loss_type (#1653)

  Show description for b344bce
  Abilityguy

  authored
  b344bceView commit details

  Copy full SHA for b344bce

  Browse repository at this point

• Fix inheritance order in PPOv2Config (#1659)

  Show description for 35e12dc
  Nicolinho

  authored
  35e12dcView commit details

  Copy full SHA for 35e12dc

  Browse repository at this point

• docs: correct cDPO usage in DPOTrainer (#1655)

  AliBakly

  authored
  1da6be1View commit details

  Copy full SHA for 1da6be1

  Browse repository at this point

• add support for training collator (#1658)

  younesbelkada

  authored
  e249cd8View commit details

  Copy full SHA for e249cd8

  Browse repository at this point

• Apply deprecated `evaluation_strategy` (#1559)

  Show description for a02513c
  

  muellerzr

  and

  kashif

  authored
  a02513cView commit details

  Copy full SHA for a02513c

  Browse repository at this point

Commits on May 22, 2024

• PPO / Reinforce Trainers (#1540)

  Show description for 13454d2
  vwxyzjn

  authored
  13454d2View commit details

  Copy full SHA for 13454d2

  Browse repository at this point

Commits on May 15, 2024

• don't cast the trainable lora layers to half precision (#1644)

  Show description for 99f2c94
  pacman100

  authored
  99f2c94View commit details

  Copy full SHA for 99f2c94

  Browse repository at this point

Commits on May 14, 2024

• Pairwise Noise Contrastive Alignment (#1632)

  Show description for 6401d08
  

  winglian

  and

  younesbelkada

  authored
  6401d08View commit details

  Copy full SHA for 6401d08

  Browse repository at this point

• Fixed wrong logs prefixes in KTOTrainer (#1641)

  Show description for d632a5b
  bartoszzuk

  authored
  d632a5bView commit details

  Copy full SHA for d632a5b

  Browse repository at this point

Commits on May 10, 2024

• Update sft_llama2.py to work with the latest API (#1637)

  Show description for 5aeb752
  xianbaoqian

  authored
  5aeb752View commit details

  Copy full SHA for 5aeb752

  Browse repository at this point

• [ORPO] Correct label mask for pad tokens (#1625)

  Show description for b8b8978
  

  IlyaGusev

  and

  kashif

  authored
  b8b8978View commit details

  Copy full SHA for b8b8978

  Browse repository at this point

• visualize rm prediction (#1636)

  Show description for 8799952
  vwxyzjn

  authored
  8799952View commit details

  Copy full SHA for 8799952

  Browse repository at this point