Commits

Branch selector

main

User selector

All users

All time

Commit history

Commits on Mar 26, 2025

• 💎 Gemma 3 VLM SFT example script for single-image and multi-image (#3131)

  Show description for 26d8675
  

  sergiopaniego

  and

  qgallouedec

  authored
  26d8675View commit details

  Copy full SHA for 26d8675

  Browse repository at this point

• 💰 Richer rich table - log all the rewards (#3156)

  qgallouedec

  authored
  9771f25View commit details

  Copy full SHA for 9771f25

  Browse repository at this point

• 👨‍🍳 vLLM serve: destroy process group on exit and pass `worker_cls` as string (#3159)

  qgallouedec

  authored
  7bdedd4View commit details

  Copy full SHA for 7bdedd4

  Browse repository at this point

Commits on Mar 25, 2025

• 🔫 Disable triggering CI when PR is draft (#3154)

  qgallouedec

  authored
  a069a2fView commit details

  Copy full SHA for a069a2f

  Browse repository at this point

• ⚰️ Remove deprecated (#3153)

  qgallouedec

  authored
  ea45f51View commit details

  Copy full SHA for ea45f51

  Browse repository at this point

• 🩹 Fix CI (#3155)

  qgallouedec

  authored
  a910239View commit details

  Copy full SHA for a910239

  Browse repository at this point

• Enable number of printed completions to be set (#3149)

  Show description for 1a9387b
  

  lewtun

  and

  qgallouedec

  authored
  1a9387bView commit details

  Copy full SHA for 1a9387b

  Browse repository at this point

Commits on Mar 24, 2025

• 🤝 Align GRPO equation doc with the implementation (#3151)

  qgallouedec

  authored
  1884ff1View commit details

  Copy full SHA for 1884ff1

  Browse repository at this point

• 🐇 [Research] Layer Skip SFT (#3111)

  Show description for bfe2075
  

  3 peopleauthored
  bfe2075View commit details

  Copy full SHA for bfe2075

  Browse repository at this point

• BCOTrainer version upgrade fixes (#2867)

  Show description for 6067e2a
  

  claralp

  and

  Clara Luise Pohland

  authored
  6067e2aView commit details

  Copy full SHA for 6067e2a

  Browse repository at this point

Commits on Mar 23, 2025

• 📊 Fix `clip_ratio` logging and better document logged values (#3145)

  qgallouedec

  authored
  dee3734View commit details

  Copy full SHA for dee3734

  Browse repository at this point

• Fix: Multi gpu hang for ORPO and CPO Trainer (#3069)

  NanoCode012

  authored
  8037f18View commit details

  Copy full SHA for 8037f18

  Browse repository at this point

Commits on Mar 22, 2025

• ⬆️ Bump dev version

  qgallouedec

  committed
  a0a5317View commit details

  Copy full SHA for a0a5317

  Browse repository at this point

• Release: v0.16 (#3137)

  qgallouedec

  authored
  23a635eView commit details

  Copy full SHA for 23a635e

  Browse repository at this point

• ⚖️ Add option not to scale rewards (Dr. GRPO) (#3135)

  qgallouedec

  authored
  9b38b0bView commit details

  Copy full SHA for 9b38b0b

  Browse repository at this point

• ☎️ Documentation for disable gathering of model weights for generation in DeepSpeed ZeRO-3 (#3136)

  qgallouedec

  authored
  0f26049View commit details

  Copy full SHA for 0f26049

  Browse repository at this point

• ⚡ Pack 300 times faster, truncate 100 times faster (#3009)

  Show description for 7511aa4
  

  mariosasko

  and

  qgallouedec

  authored
  7511aa4View commit details

  Copy full SHA for 7511aa4

  Browse repository at this point

Commits on Mar 21, 2025

• 🚀 Scaling GRPO to 70B+ Models and Multi-Node Training with vLLM Server & NCCL Communication (#3094)

  Show description for f713f61
  

  4 peopleauthored
  f713f61View commit details

  Copy full SHA for f713f61

  Browse repository at this point

Commits on Mar 20, 2025

• 🎬 Clip higher (#3118)

  Show description for a349879
  

  shirinyamani

  and

  qgallouedec

  authored
  a349879View commit details

  Copy full SHA for a349879

  Browse repository at this point

Commits on Mar 18, 2025

• Merge pull request #3079 from huggingface/flexible_reward

  Show description for 0f88c17
  shirinyamani

  authored
  0f88c17View commit details

  Copy full SHA for 0f88c17

  Browse repository at this point

• Use main process for dataset.map (#3106)

  lewtun

  authored
  beda432View commit details

  Copy full SHA for beda432

  Browse repository at this point

Commits on Mar 17, 2025

• add "_prepare_fsdp" for DPOTrainer (#2539)

  Show description for 07cfe16
  

  faaany

  and

  lewtun

  authored
  07cfe16View commit details

  Copy full SHA for 07cfe16

  Browse repository at this point

Commits on Mar 15, 2025

• 🕊️ Padding-free for SFT (#3076)

  qgallouedec

  authored
  9f7755dView commit details

  Copy full SHA for 9f7755d

  Browse repository at this point

• Update grpo_trainer.md [ci skip]

  qgallouedec

  authored
  4e3f569View commit details

  Copy full SHA for 4e3f569

  Browse repository at this point

• title multi-task added for example4

  shirinyamani

  committed
  979fda1View commit details

  Copy full SHA for 979fda1

  Browse repository at this point

• precommit fixed applied

  shirinyamani

  committed
  f6fb6a8View commit details

  Copy full SHA for f6fb6a8

  Browse repository at this point

• Merge branch 'flexible_reward' of github.com:huggingface/trl into flexible_reward

  shirinyamani

  committed
  6cbf8fbView commit details

  Copy full SHA for 6cbf8fb

  Browse repository at this point

• ⛔ Add EOS token to processed input in SFT (#3091)

  Show description for 5cb390c
  qgallouedec

  authored
  5cb390cView commit details

  Copy full SHA for 5cb390c

  Browse repository at this point

• Update trl/trainer/grpo_trainer.py

  Show description for b3c391e
  

  shirinyamani

  and

  qgallouedec

  authored
  b3c391eView commit details

  Copy full SHA for b3c391e

  Browse repository at this point

• grpo doc updated

  shirinyamani

  committed
  1b85ca6View commit details

  Copy full SHA for 1b85ca6

  Browse repository at this point

• Update docs/source/grpo_trainer.md

  Show description for e7a1290
  

  shirinyamani

  and

  qgallouedec

  authored
  e7a1290View commit details

  Copy full SHA for e7a1290

  Browse repository at this point

• Update docs/source/grpo_trainer.md

  Show description for 3822edd
  

  shirinyamani

  and

  qgallouedec

  authored
  3822eddView commit details

  Copy full SHA for 3822edd

  Browse repository at this point

• Update docs/source/grpo_trainer.md

  Show description for 230455c
  

  shirinyamani

  and

  qgallouedec

  authored
  230455cView commit details

  Copy full SHA for 230455c

  Browse repository at this point

• Update trl/trainer/grpo_trainer.py

  Show description for 08f014d
  

  shirinyamani

  and

  qgallouedec

  authored
  08f014dView commit details

  Copy full SHA for 08f014d

  Browse repository at this point

• Update trl/trainer/grpo_trainer.py

  Show description for 1074033
  

  shirinyamani

  and

  qgallouedec

  authored
  1074033View commit details

  Copy full SHA for 1074033

  Browse repository at this point