Commits

Breadcrumbs

History for

onmain

User selector

All users

All time

Commit history

Commits on Jun 14, 2025

• Enable testing during training

  ruiyiw

  committed
  dfc10b4View commit details

  Copy full SHA for dfc10b4

  View code at this pointBrowse repository at this point

Commits on Jun 10, 2025

• Debugging multiturn PPO

  ruiyiw

  committed
  ab13e6aView commit details

  Copy full SHA for ab13e6a

  View code at this pointBrowse repository at this point

Commits on Jun 9, 2025

• Save checkpoints, convert to safetensors, and upload to HF repo

  ruiyiw

  committed
  f2f0935View commit details

  Copy full SHA for f2f0935

  View code at this pointBrowse repository at this point

Commits on Jun 5, 2025

• Debug multiturn ppo

  ruiyiw

  committed
  414e9abView commit details

  Copy full SHA for 414e9ab

  View code at this pointBrowse repository at this point

Commits on Jun 3, 2025

• Enable loss mask for multiturn rollout

  ruiyiw

  committed
  f52ca2eView commit details

  Copy full SHA for f52ca2e

  View code at this pointBrowse repository at this point

• Fix but in input_ids and masks

  ruiyiw

  committed
  8aabaaeView commit details

  Copy full SHA for 8aabaae

  View code at this pointBrowse repository at this point

Commits on Jun 2, 2025

• Resolve conflict

  ruiyiw

  committed
  4f03865View commit details

  Copy full SHA for 4f03865

  View code at this pointBrowse repository at this point

• [algo] OPO: add implementations and descriptions for OPO (On-Policy RL with Optimal Reward Baseline) (#1796)

  Show description for a4b1bb7
  

  YRdddream

  and

  eric-haibin-lin

  authored
  a4b1bb7View commit details

  Copy full SHA for a4b1bb7

  View code at this pointBrowse repository at this point

• [AMD] fix: Add support for RAY_EXPERIMENTAL_NOSET_*_VISIBLE_DEVICES (Fix AMD support) (#1465)

  Show description for 07897f8
  HollowMan6

  authored
  07897f8View commit details

  Copy full SHA for 07897f8

  View code at this pointBrowse repository at this point

• [bugfix] fix select_idxs function in DataProto (#1794)

  Show description for ea81658
  

  Moyu-42

  and

  keithsun

  authored
  ea81658View commit details

  Copy full SHA for ea81658

  View code at this pointBrowse repository at this point

• chore: fix typos across codebase (#1805)

  Show description for 0e127b2
  sukrucildirr

  authored
  0e127b2View commit details

  Copy full SHA for 0e127b2

  View code at this pointBrowse repository at this point

• [feat] worker_group: support custom label for specific devices (#1773)

  Show description for 6e0e860
  ZDJeffrey

  authored
  6e0e860View commit details

  Copy full SHA for 6e0e860

  View code at this pointBrowse repository at this point

Commits on Jun 1, 2025

• remove redundant 'get_custom_reward_fn' function (#1791)

  Show description for 3126c8b
  vllbc

  authored
  3126c8bView commit details

  Copy full SHA for 3126c8b

  View code at this pointBrowse repository at this point

• fix import issue for mcore package (#1775)

  Show description for 1fd729c
  

  jinqinn

  authored
  1fd729cView commit details

  Copy full SHA for 1fd729c

  View code at this pointBrowse repository at this point

• fix freeze router weights for Qwen2MoE (#1792)

  yzlnew

  authored
  ad94700View commit details

  Copy full SHA for ad94700

  View code at this pointBrowse repository at this point

• [doc] fix: Fix `doc_test`ci workflow pipeline (#1767)

  Show description for 0ae5056
  hongpeng-guo

  authored
  0ae5056View commit details

  Copy full SHA for 0ae5056

  View code at this pointBrowse repository at this point

• [sglang] refactor: Unify async rollout under SGLangRollout, and support sglang==0.4.6.post5 (#1717)

  Show description for 4de247f
  

  5 peopleauthored
  4de247fView commit details

  Copy full SHA for 4de247f

  View code at this pointBrowse repository at this point

Commits on May 31, 2025

• [docs] ppo: add a page for PPO algorithm (#1781)

  Show description for 106d33f
  eric-haibin-lin

  authored
  106d33fView commit details

  Copy full SHA for 106d33f

  View code at this pointBrowse repository at this point

• [sglang] Feat: Search Tool Invocation in Multi-Turn RL Training (#1682)

  Lins-01

  authored
  c5bc81bView commit details

  Copy full SHA for c5bc81b

  View code at this pointBrowse repository at this point

Commits on May 30, 2025

• Fix minor bug

  ruiyiw

  committed
  6c80129View commit details

  Copy full SHA for 6c80129

  View code at this pointBrowse repository at this point

• Fix minor bug

  ruiyiw

  committed
  fcbbc2dView commit details

  Copy full SHA for fcbbc2d

  View code at this pointBrowse repository at this point

• Fix bug of prompt/response_ids larger than max length

  ruiyiw

  committed
  4a9a4a4View commit details

  Copy full SHA for 4a9a4a4

  View code at this pointBrowse repository at this point

• Fix error of non_tensor_batch concatenation

  ruiyiw

  committed
  6bee540View commit details

  Copy full SHA for 6bee540

  View code at this pointBrowse repository at this point

• Fix bug of prompt/response_ids larger than max length

  ruiyiw

  committed
  982e72eView commit details

  Copy full SHA for 982e72e

  View code at this pointBrowse repository at this point

• [feat] dataproto: Supporting new operations (sample_level_repeat, unfold_column_chunks) for `DataProto` (#1761)

  Show description for e23e67b
  hongpeng-guo

  authored
  e23e67bView commit details

  Copy full SHA for e23e67b

  View code at this pointBrowse repository at this point

• [refactor] update func generator implementation to improve its observability (#1762)

  Show description for ed3767d
  hongpeng-guo

  authored
  ed3767dView commit details

  Copy full SHA for ed3767d

  View code at this pointBrowse repository at this point

• fix a bug that Moe's GPU memory offload is not properly handled (#1766)

  Show description for 2981aa2
  LiuXTao

  authored
  2981aa2View commit details

  Copy full SHA for 2981aa2

  View code at this pointBrowse repository at this point

• [fix] self.reward_module._handle.reshard(True) not required for fsdp2 (#1765)

  Show description for 28a31d1
  liyc-ai

  authored
  28a31d1View commit details

  Copy full SHA for 28a31d1

  View code at this pointBrowse repository at this point

• [feat] support kimi_vl VLM model (#1639)

  Show description for 96903e0
  ShareLer

  authored
  96903e0View commit details

  Copy full SHA for 96903e0

  View code at this pointBrowse repository at this point

• [vlm] Support ulysses sequence parallelism for vlm (#1739)

  Show description for 9c50ffd
  popomen

  authored
  9c50ffdView commit details

  Copy full SHA for 9c50ffd

  View code at this pointBrowse repository at this point

• Adapt to changes in scripts/model_merger.py

  ruiyiw

  committed
  f3ed06cView commit details

  Copy full SHA for f3ed06c

  View code at this pointBrowse repository at this point

• Add validation metrics for interactive multi-turn dense reward RL

  ruiyiw

  committed
  4f92641View commit details

  Copy full SHA for 4f92641

  View code at this pointBrowse repository at this point

Commits on May 29, 2025

• Resolve conflict

  ruiyiw

  committed
  e012d02View commit details

  Copy full SHA for e012d02

  View code at this pointBrowse repository at this point

• Support dense reward assignment for multiturn RL

  ruiyiw

  committed
  2244b6cView commit details

  Copy full SHA for 2244b6c

  View code at this pointBrowse repository at this point

• [fix] moonlight runnable with trust_remote_code (#1749)

  ETOgaosion

  authored
  55f13ffView commit details

  Copy full SHA for 55f13ff

  View code at this pointBrowse repository at this point