Commits

Branch selector

main

User selector

All users

All time

Commit history

Commits on Jan 27, 2024

• Update dpo_trainer.mdx (#1280)

  ouhenio

  authored
  ef441eaView commit details

  Copy full SHA for ef441ea

  Browse repository at this point

Commits on Jan 26, 2024

• Fix sft ci (#1279)

  vwxyzjn

  authored
  af623aeView commit details

  Copy full SHA for af623ae

  Browse repository at this point

• Fix SFT tuner (#1278)

  vwxyzjn

  authored
  3843cfcView commit details

  Copy full SHA for 3843cfc

  Browse repository at this point

• Remove tyro (#1176)

  Show description for 9a71e67
  

  3 peopleauthored
  9a71e67View commit details

  Copy full SHA for 9a71e67

  Browse repository at this point

• FIx SFTTrainer bugs on TRL main (#1276)

  Show description for 09ca565
  younesbelkada

  authored
  09ca565View commit details

  Copy full SHA for 09ca565

  Browse repository at this point

• Only load data on main process (#1255)

  Show description for 4edc688
  

  JohnGiorgi

  and

  younesbelkada

  authored
  4edc688View commit details

  Copy full SHA for 4edc688

  Browse repository at this point

Commits on Jan 24, 2024

• [DPO] average_log_prob when loss is IPO (#1265)

  Show description for 29d439a
  kashif

  authored
  29d439aView commit details

  Copy full SHA for 29d439a

  Browse repository at this point

Commits on Jan 23, 2024

• Fix typo in extra_columns variable name (#1269)

  Show description for 5760e5d
  otlaitil

  authored
  5760e5dView commit details

  Copy full SHA for 5760e5d

  Browse repository at this point

Commits on Jan 22, 2024

• Update utils.py (#1256)

  philschmid

  authored
  a3c5b71View commit details

  Copy full SHA for a3c5b71

  Browse repository at this point

Commits on Jan 19, 2024

• set dev version (#1254)

  younesbelkada

  authored
  222d275View commit details

  Copy full SHA for 222d275

  Browse repository at this point

• Release: v0.7.10 (#1253)

  younesbelkada

  authored
  09ca760View commit details

  Copy full SHA for 09ca760

  Browse repository at this point

Commits on Jan 18, 2024

• fix: fix loss_type and some args desc (#1247)

  zspo

  authored
  1e68753View commit details

  Copy full SHA for 1e68753

  Browse repository at this point

• Fix chatml template (#1248)

  Show description for 1f59eeb
  

  philschmid

  and

  younesbelkada

  authored
  1f59eebView commit details

  Copy full SHA for 1f59eeb

  Browse repository at this point

• Add `setup_chat_format` for adding new special tokens to model for training chat models (#1242)

  Show description for 928d144
  

  philschmid

  and

  younesbelkada

  authored
  928d144View commit details

  Copy full SHA for 928d144

  Browse repository at this point

• Fix weird doc bug (#1244)

  Show description for 3319993
  younesbelkada

  authored
  3319993View commit details

  Copy full SHA for 3319993

  Browse repository at this point

Commits on Jan 17, 2024

• Update sft_trainer.py (#1241)

  younesbelkada

  authored
  4fb3d0cView commit details

  Copy full SHA for 4fb3d0c

  Browse repository at this point

• [`core` / SFTTrainer] Fix breaking change (#1229)

  Show description for bcccdeb
  younesbelkada

  authored
  bcccdebView commit details

  Copy full SHA for bcccdeb

  Browse repository at this point

• [`core` / tests ] v1 slow tests (#1218)

  Show description for ef209e3
  younesbelkada

  authored
  ef209e3View commit details

  Copy full SHA for ef209e3

  Browse repository at this point

• fix: improve error message when `pad_token_id` is not configured (#1152)

  Show description for 341f6a6
  yumemio

  authored
  341f6a6View commit details

  Copy full SHA for 341f6a6

  Browse repository at this point

Commits on Jan 15, 2024

• Update dpo_trainer.py (#1160)

  Show description for 97b9fa2
  AjayP13

  authored
  97b9fa2View commit details

  Copy full SHA for 97b9fa2

  Browse repository at this point

• Remove a repeating line in how_to_train.md (#1226)

  kykim0

  authored
  a7d796cView commit details

  Copy full SHA for a7d796c

  Browse repository at this point

Commits on Jan 12, 2024

• Create slow-tests.yml (#1223)

  younesbelkada

  authored
  fa074e6View commit details

  Copy full SHA for fa074e6

  Browse repository at this point

• Add support for ChatML dataset format in (#1208)

  Show description for 776939d
  philschmid

  authored
  776939dView commit details

  Copy full SHA for 776939d

  Browse repository at this point

• Refactor RewardConfig to own module (#1221)

  Show description for 163ca9f
  lewtun

  authored
  163ca9fView commit details

  Copy full SHA for 163ca9f

  Browse repository at this point

Commits on Jan 11, 2024

• [`core` / `Docker`] Add workflow to build TRL docker images (#1215)

  Show description for 2eeb7b0
  younesbelkada

  authored
  2eeb7b0View commit details

  Copy full SHA for 2eeb7b0

  Browse repository at this point

Commits on Jan 10, 2024

• Fix args type (#1214)

  Show description for 9f8d0e4
  zspo

  authored
  9f8d0e4View commit details

  Copy full SHA for 9f8d0e4

  Browse repository at this point

• Update Unsloth SFT, DPO docs (#1213)

  Show description for c9b7145
  danielhanchen

  authored
  c9b7145View commit details

  Copy full SHA for c9b7145

  Browse repository at this point

Commits on Jan 9, 2024

• Fix FSDP error (#1196)

  Show description for baf3c1c
  

  mgerstgrasser

  and

  younesbelkada

  authored
  baf3c1cView commit details

  Copy full SHA for baf3c1c

  Browse repository at this point

• Fix shape descriptions in calculate_loss method (#1204)

  yuta0x89

  authored
  b181e40View commit details

  Copy full SHA for b181e40

  Browse repository at this point

• Check tokenize params on DPOTrainer (#1197)

  Show description for 26da9e8
  pablovicente

  authored
  26da9e8View commit details

  Copy full SHA for 26da9e8

  Browse repository at this point

• set dev version (#1207)

  younesbelkada

  authored
  d6cc88aView commit details

  Copy full SHA for d6cc88a

  Browse repository at this point

• release: v0.7.9 (#1206)

  younesbelkada

  authored
  7a95cc8View commit details

  Copy full SHA for 7a95cc8

  Browse repository at this point

• Revert "Address issue #1122 (#1174)" (#1205)

  Show description for d171551
  younesbelkada

  authored
  d171551View commit details

  Copy full SHA for d171551

  Browse repository at this point

• [`DPOTrainer`] Fix peft + DPO + bf16 if one uses `generate_during_eval` or pre-computed logits (#1203)

  Show description for d116887
  younesbelkada

  authored
  d116887View commit details

  Copy full SHA for d116887

  Browse repository at this point

• Fix reported KL in PPO trainer (#1180)

  Show description for a236c57
  mgerstgrasser

  authored
  a236c57View commit details

  Copy full SHA for a236c57

  Browse repository at this point