Code

🔁 🦈 Support iterative GRPO #2700

Code

Merged

qgallouedec merged 15 commits intohuggingface:mainhuggingface/trl:mainfrom shirinyamani:mainshirinyamani/trl:mainCopy head branch name to clipboardFeb 4, 2025

+75

Commits

Commits on Jan 30, 2025

• support for synchronization ref-model added

  shirinyamani

  committed
  e047b80

  Copy full SHA for e047b80

• support for synchronization ref-model added

  shirinyamani

  committed
  aa4732e

  Copy full SHA for aa4732e

• Merge branch 'huggingface:main' into main

  shirinyamani

  authored
  c49aa19

  Copy full SHA for c49aa19

• tests for sync_ref_model added

  shirinyamani

  committed
  1fd9c71

  Copy full SHA for 1fd9c71

• Merge branch 'main' of github.com:shirinyamani/trl

  shirinyamani

  committed
  2ad4ac4

  Copy full SHA for 2ad4ac4

• Update tests/test_grpo_trainer.py

  Show description for dda1a82
  

  shirinyamani

  and

  qgallouedec

  authored
  dda1a82

  Copy full SHA for dda1a82

Commits on Feb 3, 2025

• Merge branch 'huggingface:main' into main

  shirinyamani

  authored
  e631a9b

  Copy full SHA for e631a9b

Commits on Feb 4, 2025

• split and fix test

  qgallouedec

  committed
  cb393f5

  Copy full SHA for cb393f5

• style

  qgallouedec

  committed
  f2026a9

  Copy full SHA for f2026a9

• doc

  qgallouedec

  committed
  0b38b54

  Copy full SHA for 0b38b54

• move after init to ensure accelerator exists

  qgallouedec

  committed
  71b0fc4

  Copy full SHA for 71b0fc4

• Merge branch 'main' into main

  qgallouedec

  authored
  1f865a4

  Copy full SHA for 1f865a4

• Update tests/test_grpo_trainer.py

  qgallouedec

  authored
  800e9ee

  Copy full SHA for 800e9ee

• style

  qgallouedec

  committed
  f828e7c

  Copy full SHA for f828e7c

• Merge branch 'main' into main

  qgallouedec

  authored
  4ced3a1

  Copy full SHA for 4ced3a1