Commits

Breadcrumbs

History for

onmain

User selector

All users

All time

Commit history

Commits on Feb 14, 2026

• [fsdp,algo] feat: add NVFP4 QAT (Quantization-Aware Training) support (#5190)

  Show description for 27f65f1
  zhangyimi

  authored
  27f65f1View commit details

  Copy full SHA for 27f65f1

  View code at this pointBrowse repository at this point

Commits on Jan 29, 2026

• [megatron] feat: Support MTP training in SFT (#4981)

  Show description for adff795
  

  3 peopleauthored
  adff795View commit details

  Copy full SHA for adff795

  View code at this pointBrowse repository at this point

Commits on Jan 23, 2026

• [BREAKING][worker, rollout, vllm] feat: implement vLLM colocated training-inference rollout with process separation (#4280)

  Show description for e9405d7
  

  jianjunzhong

  and

  wuxibin89

  authored
  e9405d7View commit details

  Copy full SHA for e9405d7

  View code at this pointBrowse repository at this point

Commits on Jan 21, 2026

• [perf] feat: verl profiler system support Agent Loop scenario and integrate torch.profiler (#4320)

  Show description for cd1926c
  

  mengchengTang

  and

  tardis-key

  authored
  cd1926cView commit details

  Copy full SHA for cd1926c

  View code at this pointBrowse repository at this point

Commits on Jan 19, 2026

• [megatron] feat: Using MTP in RL Training and Inference (#4936)

  Show description for df340d7
  ArronHZG

  authored
  df340d7View commit details

  Copy full SHA for df340d7

  View code at this pointBrowse repository at this point

Commits on Jan 7, 2026

• [algo] feat: add optimal token baseline and variance proxy (#4678)

  Show description for 67a127b
  

  3 peopleauthored
  67a127bView commit details

  Copy full SHA for 67a127b

  View code at this pointBrowse repository at this point

Commits on Jan 6, 2026

• [recipe] feat: migrate `recipe` to the dedicated repo `verl-recipe` as a submodule (#4795)

  Show description for 2bb42ba
  tongyx361

  authored
  2bb42baView commit details

  Copy full SHA for 2bb42ba

  View code at this pointBrowse repository at this point

Commits on Jan 5, 2026

• [fsdp] feat: integrate PrefixGrouper for GRPO training acceleration (#4368)

  Show description for d6bc629
  kevssim

  authored
  d6bc629View commit details

  Copy full SHA for d6bc629

  View code at this pointBrowse repository at this point

Commits on Jan 2, 2026

• Revert "[algo] fix: Add seq mean mask denominator option" (#4769)

  Show description for 78014a2
  wuxibin89

  authored
  78014a2View commit details

  Copy full SHA for 78014a2

  View code at this pointBrowse repository at this point

Commits on Dec 30, 2025

• [training_utils] fix: RM extra scaling in KL/PG losses (#4711)

  Show description for 0af6a38
  JacobHelwig

  authored
  0af6a38View commit details

  Copy full SHA for 0af6a38

  View code at this pointBrowse repository at this point

Commits on Dec 29, 2025

• [megatron] fix: set model to eval during compute_log_prob/compute_values (#4708)

  Show description for c790552
  HollowMan6

  authored
  c790552View commit details

  Copy full SHA for c790552

  View code at this pointBrowse repository at this point

Commits on Dec 17, 2025

• [algo] fix: Add seq mean mask denominator option (#4510)

  Show description for 6a58521
  

  szrlee

  and

  tongyx361

  authored
  6a58521View commit details

  Copy full SHA for 6a58521

  View code at this pointBrowse repository at this point

Commits on Dec 15, 2025

• [algo,doc] refactor: rollout correction (#4511)

  Show description for 518bada
  szrlee

  authored
  518badaView commit details

  Copy full SHA for 518bada

  View code at this pointBrowse repository at this point

Commits on Dec 4, 2025

• [algo] feat: support router replay (#4101)

  Show description for cb23607
  

  3 peopleauthored
  cb23607View commit details

  Copy full SHA for cb23607

  View code at this pointBrowse repository at this point

Commits on Dec 3, 2025

• [fsdp] fix: Fixing the error caused by empty tensors in the multi_turn + remove_padding scenario (#4165)

  Show description for 80c860f
  nuerxiati

  authored
  80c860fView commit details

  Copy full SHA for 80c860f

  View code at this pointBrowse repository at this point

Commits on Dec 2, 2025

• [megatron] feat: support gpt-oss (#4323)

  Show description for 9f3199f
  ISEEKYAN

  authored
  9f3199fView commit details

  Copy full SHA for 9f3199f

  View code at this pointBrowse repository at this point

Commits on Nov 25, 2025

• [misc] feat: Add `actor_rollout_ref.actor.calculate_entropy` for entropy fwd (#4239)

  Show description for bae6f35
  EduardDurech

  authored
  bae6f35View commit details

  Copy full SHA for bae6f35

  View code at this pointBrowse repository at this point

Commits on Nov 13, 2025

• [worker, trainer, recipe] feat: add FP16 training and inference support (#4036)

  Show description for 65f9bad
  

  3 peopleauthored
  65f9badView commit details

  Copy full SHA for 65f9bad

  View code at this pointBrowse repository at this point

Commits on Nov 12, 2025

• [doc,algo] feat: Rollout Correction - Fix Metrics, Add Documentation, and Add Batch Normalization (#4070)

  Show description for 2c6c65c
  

  szrlee

  and

  tongyx361

  authored
  2c6c65cView commit details

  Copy full SHA for 2c6c65c

  View code at this pointBrowse repository at this point

Commits on Nov 10, 2025

• [algo] feat: return loss and metrics from policy_loss_fn (#4062)

  Show description for a7f58cd
  

  tongyx361

  and

  gemini-code-assist[bot]

  authored
  a7f58cdView commit details

  Copy full SHA for a7f58cd

  View code at this pointBrowse repository at this point

Commits on Nov 7, 2025

• [BREAKING][algo] feat: Rollout Correction for General Off-Policy Problems (#3984)

  Show description for 418f964
  

  szrlee

  and

  tongyx361

  authored
  418f964View commit details

  Copy full SHA for 418f964

  View code at this pointBrowse repository at this point

Commits on Nov 6, 2025

• [env] feat: update docker file building schema, from VLLM base images (#3937)

  Show description for 77814f1
  

  ISEEKYAN

  and

  wuxibin89

  authored
  77814f1View commit details

  Copy full SHA for 77814f1

  View code at this pointBrowse repository at this point

Commits on Oct 30, 2025

• [Megatron] feat: 1f1b overlap/moe_a2a_overlap (#3522)

  Show description for 576fbcc
  ISEEKYAN

  authored
  576fbccView commit details

  Copy full SHA for 576fbcc

  View code at this pointBrowse repository at this point

Commits on Oct 24, 2025

• [megatron] fix: VLMs using fused kernels (#3849)

  Show description for ab676dc
  HollowMan6

  authored
  ab676dcView commit details

  Copy full SHA for ab676dc

  View code at this pointBrowse repository at this point

Commits on Oct 17, 2025

• [trainer, recipe] feat: fully async training recipe (#2981)

  Show description for b25bb7d
  

  7 peopleauthored
  b25bb7dView commit details

  Copy full SHA for b25bb7d

  View code at this pointBrowse repository at this point

Commits on Oct 13, 2025

• [BREAKING][rollout, trainer, algo] feat: comprehensive rollout importance sampling implementation (#3694)

  Show description for 21271aa
  

  szrlee

  and

  ISEEKYAN

  authored
  21271aaView commit details

  Copy full SHA for 21271aa

  View code at this pointBrowse repository at this point

Commits on Sep 28, 2025

• [megatron, worker] fix: use `extract_multi_modal_inputs` method for handling `multi_modal_inputs` (#3641)

  Show description for abca659
  HollowMan6

  authored
  abca659View commit details

  Copy full SHA for abca659

  View code at this pointBrowse repository at this point

Commits on Sep 26, 2025

• [trainer] fix: Ref to #3596. More import fix for transformers version higher than 4.55.0 (#3608)

  Show description for 64a9860
  A1waysBeenHere

  authored
  64a9860View commit details

  Copy full SHA for 64a9860

  View code at this pointBrowse repository at this point

Commits on Sep 25, 2025

• [worker] fix: correctly determine is_vlm_model if sp > 1 (#3282)

  Show description for 616e933
  HollowMan6

  authored
  616e933View commit details

  Copy full SHA for 616e933

  View code at this pointBrowse repository at this point

Commits on Sep 24, 2025

• [trainer] fix: Import flash attn utils for Transformers higher than 4.55.0 (#3596)

  Show description for ba85551
  

  3 peopleauthored
  ba85551View commit details

  Copy full SHA for ba85551

  View code at this pointBrowse repository at this point

• [misc] feat: prototype deprecate DataProto and replace with Tensordict: part 2 (#3567)

  Show description for 69b0127
  

  houminz

  and

  vermouth1992

  authored
  69b0127View commit details

  Copy full SHA for 69b0127

  View code at this pointBrowse repository at this point

• [megatron] fix: revert megatron actor refactor (#3553)

  Show description for 1985eb1
  vermouth1992

  authored
  1985eb1View commit details

  Copy full SHA for 1985eb1

  View code at this pointBrowse repository at this point

Commits on Sep 19, 2025

• [worker] fix: get all `multi_modal_inputs` keys with in a microbatch (#3315)

  Show description for 2f6a5d6
  HollowMan6

  authored
  2f6a5d6View commit details

  Copy full SHA for 2f6a5d6

  View code at this pointBrowse repository at this point

Commits on Sep 18, 2025

• [megatron] Add TIS support to megatron backend (#3513)

  Show description for b6b34b2
  

  sharonyu-115

  and

  Shuang Yu

  authored
  b6b34b2View commit details

  Copy full SHA for b6b34b2

  View code at this pointBrowse repository at this point

• [model] fix: refactor qwen2vl patches & support no-image input for fsdp (#3496)

  Show description for 0d4541f
  hiyouga

  authored
  0d4541fView commit details

  Copy full SHA for 0d4541f

  View code at this pointBrowse repository at this point