Commits

Branch selector

 bb56c6e

User selector

All users

Datepicker

All time

Commit History

Commits on Oct 31, 2024

• 💾 Fix _save_checkpoint for online methods (#2288)

  qgallouedec

  authored
  bb56c6e
  More actions

• 🖇️ Better dependency and partitioning of CI tests (#2298)

  qgallouedec

  authored
  06be6f4
  More actions

Commits on Oct 28, 2024

• 🍬 Use any reward model for online methods (#2276)

  qgallouedec

  authored
  b269657
  More actions

• 🔌 Fix type hint in LogCompletionsCallback (#2285)

  

  anch0vy

  and

  qgallouedec

  authored
  0ce3b65
  More actions

• ⛓️‍💥 Don't use eval_dataset in scripts when no eval strategy (#2270)

  qgallouedec

  authored
  e155cb8
  More actions

Commits on Oct 25, 2024

• 🧮 Fix the computation of KL divergence loss (#2277)

  d-tiapkin

  authored
  ea7a1be
  More actions

• 🏁 Add bos_token_id only if it exists (#2279)

  

  seanexp

  and

  sean.jung

  authored
  110d088
  More actions

Commits on Oct 24, 2024

• 🧘 Replace F.log(F.sigmoid(log_odds) with F.logsigmoid(log_odds) (#2274)

  

  zhanwenchen

  and

  qgallouedec

  authored
  57ba9b9
  More actions

• 🧼 Refactor log_reports.py for Improved Logging, File Processing, and Slack Payload Handling (#2249)

  

  Mefisto04

  and

  qgallouedec

  authored
  0de75b2
  More actions

• ♾️ Fix test generation max_new_tokens (#2272)

  qgallouedec

  authored
  e615974
  More actions

• Add torch_dtype to model kwargs in reward modeling example (#2266)

  

  cameronphchen

  and

  qgallouedec

  authored
  c2bb1ee
  More actions

• [Judges] use the pair-judges in online-preference trainers (#2243)

  

  kashif

  and

  qgallouedec

  authored
  9c376c5
  More actions

• Conversational dataset support for KTOTrainer (#2248)

  

  qgallouedec

  and

  lewtun

  authored
  1699473
  More actions

• Bump the minimum transformers version to v4.46 (#2245)

  

  qgallouedec

  and

  kashif

  authored
  99225bb
  More actions

Commits on Oct 22, 2024

• 🚩 setup_chat_format: throw error if there is already a template in base model (#2252)

  

  ngxson

  and

  kashif

  authored
  88be2c0
  More actions

• Adjust padding in batch generation (#2251)

  

  gaetanlop

  and

  kashif

  authored
  f2349d2
  More actions

• Use processing_class instead of tokenizer in LogCompletionsCallback (#2261)

  qgallouedec

  authored
  d843b3d
  More actions

Commits on Oct 21, 2024

• 🧽 Fix typo in dataset format doc (#2259)

  cameronphchen

  authored
  84dab85
  More actions

• 🏗️ Refactor DPO data processing (#2209)

  

  qgallouedec

  and

  lewtun

  authored
  92f6d24
  More actions

Commits on Oct 18, 2024

• 🔀 Rename get_batch_sample and add num_items_in_batch to compute_loss (#2246)

  qgallouedec

  authored
  31b7820
  More actions

• ✨ Enhance log report script: add error handling and logging (#2232)

  

  DhruvKadam-git

  and

  qgallouedec

  authored
  b9aa965
  More actions

Commits on Oct 17, 2024

• Update SFT examples (#2244)

  lewtun

  authored
  a67f214
  More actions

• [CLI] Setting capture output to False (#2239)

  

  August-murr

  and

  qgallouedec

  authored
  494b4af
  More actions

Commits on Oct 16, 2024

• DPO support remove_unused_columns (#2233)

  qgallouedec

  authored
  02f4e75
  More actions

Commits on Oct 15, 2024

• Updated ScriptArguments warning messages (#2230)

  sergiopaniego

  authored
  2ba3005
  More actions

Commits on Oct 14, 2024

• 🎭 Deprecate [SFT/DPO/Reward]ScriptArguments in favour of ScriptArguments (#2145)

  qgallouedec

  authored
  7e394b0
  More actions

Commits on Oct 13, 2024

• ✨ Update commands for code linting in contributing guidelines (#2225)

  

  Ben-Schneider-code

  and

  qgallouedec

  authored
  14f3613
  More actions

Commits on Oct 11, 2024

• 📒 Fix type/format confusions (#2223)

  qgallouedec

  authored
  5e24101
  More actions

• Add GKD to dataset_formats.mdx (#2222)

  

  August-murr

  and

  qgallouedec

  authored
  b81a612
  More actions

• Add Sequence-Level KD (#2220)

  

  3 peopleauthored
  7f0d246
  More actions

• 🕊️ Migration PPOv2 -> PPO (#2174)

  

  3 peopleauthored
  70036bf
  More actions

• Conversational dataset support for ORPOTrainer (#2184)

  qgallouedec

  authored
  d0aa421
  More actions

• trl env report all cuda devices (#2216)

  qgallouedec

  authored
  5375d71
  More actions

• Updated README.md with CLI examples and additional usage instructions (#2199)

  

  4 peopleauthored
  6004e03
  More actions

• Update README.md (#2180)

  

  4 peopleauthored
  f436c3e
  More actions

Pagination

PreviousNext