Commits

Branch selector

main

User selector

All users

All time

Commit history

Commits on Feb 1, 2025

• [misc] fix: grpo kl loss should be add when do minimization (#179)

  Show description for a65c915
  PeterSH6

  authored
  a65c915View commit details

  Copy full SHA for a65c915

  Browse repository at this point

Commits on Jan 31, 2025

• docs: add twitter link to readme

  eric-haibin-lin

  authored
  38ac525View commit details

  Copy full SHA for 38ac525

  Browse repository at this point

• [feat, SFT] Support LigerKernel for SFT (#173)

  xingyaoww

  authored
  25fc194View commit details

  Copy full SHA for 25fc194

  Browse repository at this point

• validate `use_remove_padding` when applying sequence parallelism (#153)

  chujiezheng

  authored
  fb3793aView commit details

  Copy full SHA for fb3793a

  Browse repository at this point

• fix bug：fix checkpoint save with existing dirs (#174)

  Show description for 679798c
  

  dignfei

  and

  eric-haibin-lin

  authored
  679798cView commit details

  Copy full SHA for 679798c

  Browse repository at this point

Commits on Jan 30, 2025

• docs: add split placement to readme

  eric-haibin-lin

  authored
  b6068ecView commit details

  Copy full SHA for b6068ec

  Browse repository at this point

• Allow users to pass in custom compute_score function (#162)

  Show description for ce862ce
  corbt

  authored
  ce862ceView commit details

  Copy full SHA for ce862ce

  Browse repository at this point

• fix: typo (explanation) (#167)

  emergenz

  authored
  41b7c58View commit details

  Copy full SHA for 41b7c58

  Browse repository at this point

• fix: typo (#166)

  emergenz

  authored
  e7fd415View commit details

  Copy full SHA for e7fd415

  Browse repository at this point

• [Liger-kernel] Add an option to use `_apply_liger_kernel_to_instance()` to load model (#133)

  Show description for dd41877
  hongpeng-guo

  authored
  dd41877View commit details

  Copy full SHA for dd41877

  Browse repository at this point

• [nit] Explicitly show logits dimension in `dp_worker.py::forward_micro_batch` (#164)

  Show description for df03aa6
  hongpeng-guo

  authored
  df03aa6View commit details

  Copy full SHA for df03aa6

  Browse repository at this point

• Update README.md (#165)

  Show description for 8f6e5f5
  vermouth1992

  authored
  8f6e5f5View commit details

  Copy full SHA for 8f6e5f5

  Browse repository at this point

• [misc] fix: fix ray requirement (#163)

  vermouth1992

  authored
  29935aeView commit details

  Copy full SHA for 29935ae

  Browse repository at this point

Commits on Jan 29, 2025

• fix: redundant non_zero_mask (#152)

  Show description for f9784cd
  emergenz

  authored
  f9784cdView commit details

  Copy full SHA for f9784cd

  Browse repository at this point

• fix: fix missing version file dependency when installing without -e (#156)

  eric-haibin-lin

  authored
  e3cc0ebView commit details

  Copy full SHA for e3cc0eb

  Browse repository at this point

• docs: add ragen as a related work

  eric-haibin-lin

  authored
  33e374aView commit details

  Copy full SHA for 33e374a

  Browse repository at this point

Commits on Jan 28, 2025

• [misc]fix: pad dataproto when pad size is larger than len(dataproto) (#150)

  Show description for ab525bc
  

  PeterSH6

  and

  vermouth1992

  authored
  ab525bcView commit details

  Copy full SHA for ab525bc

  Browse repository at this point

Commits on Jan 27, 2025

• [misc] feat: enable grad ckpt as default and enable chunk prefill as default (#147)

  PeterSH6

  authored
  9fca71dView commit details

  Copy full SHA for 9fca71d

  Browse repository at this point

• [perf] docs: fix typo

  eric-haibin-lin

  authored
  54603cbView commit details

  Copy full SHA for 54603cb

  Browse repository at this point

• docs: add news for doubao-1.5-pro

  eric-haibin-lin

  authored
  b2c6ff7View commit details

  Copy full SHA for b2c6ff7

  Browse repository at this point

• Update README.md (#146)

  Show description for 12b0b59
  vermouth1992

  authored
  12b0b59View commit details

  Copy full SHA for 12b0b59

  Browse repository at this point

• [misc] fix: gradient accumulation in seq balance and modify default vllm log level (#141)

  Show description for 695bdbb
  PeterSH6

  authored
  695bdbbView commit details

  Copy full SHA for 695bdbb

  Browse repository at this point

• [doc] perf: add performance tuning guide for FSDP backend (#142)

  Show description for 78a4606
  PeterSH6

  authored
  78a4606View commit details

  Copy full SHA for 78a4606

  Browse repository at this point

• [SFT] Support context parallelism for SFT (#132)

  Show description for 077173f
  

  3 peopleauthored
  077173fView commit details

  Copy full SHA for 077173f

  Browse repository at this point

• [VLLM] Set max_num_batched_tokens for vllm rollout (#140)

  Show description for c99df03
  xingyaoww

  authored
  c99df03View commit details

  Copy full SHA for c99df03

  Browse repository at this point

• [BREAKING][misc] feat: change micro_batch_size to micro_batch_size_per_gpu (#136)

  Show description for f2a76ac
  PeterSH6

  authored
  f2a76acView commit details

  Copy full SHA for f2a76ac

  Browse repository at this point

• docs: add reference for tiny-zero

  eric-haibin-lin

  authored
  c17e6c6View commit details

  Copy full SHA for c17e6c6

  Browse repository at this point

• docs: add ray and slurm link

  eric-haibin-lin

  authored
  e954903View commit details

  Copy full SHA for e954903

  Browse repository at this point

• [misc] fix: only return old_log_prob and temp to fix union problem in old_log_prob (#137)

  Show description for 8bf9b95
  PeterSH6

  authored
  8bf9b95View commit details

  Copy full SHA for 8bf9b95

  Browse repository at this point

Commits on Jan 26, 2025

• docs: update README.md typo (#139)

  Show description for a63f4a0
  eltociear

  authored
  a63f4a0View commit details

  Copy full SHA for a63f4a0

  Browse repository at this point

• [misc] fix nan in non_tensor_batch union (#135)

  PeterSH6

  authored
  23101c3View commit details

  Copy full SHA for 23101c3

  Browse repository at this point

Commits on Jan 25, 2025

• [SFT] feat: Add LoRA support for SFT (#127)

  Show description for 6d96fda
  

  3 peopleauthored
  6d96fdaView commit details

  Copy full SHA for 6d96fda

  Browse repository at this point

Commits on Jan 24, 2025

• [readme] docs: add links for GRPO

  eric-haibin-lin

  authored
  22e9311View commit details

  Copy full SHA for 22e9311

  Browse repository at this point

• Update README.md (#130)

  vermouth1992

  authored
  a4048b4View commit details

  Copy full SHA for a4048b4

  Browse repository at this point

• [ppo] refactor: refactor old_log_prob into a separate function (#129)

  vermouth1992

  authored
  e974fb1View commit details

  Copy full SHA for e974fb1

  Browse repository at this point