Commits

Branch selector

main

User selector

All users

All time

Commit history

Commits on Dec 2, 2025

• Update tokenizer for llama3 (#144)

  Show description for de1b0ea
  

  andriigrynenko

  and

  sergei-something-ai

  authored
  de1b0eaView commit details

  Copy full SHA for de1b0ea

  Browse repository at this point

• [tinker-cookbook] Initial drop of xmux (#138)

  kennyyu

  authored
  1ce9a0fView commit details

  Copy full SHA for 1ce9a0f

  Browse repository at this point

• Load optimizer state when appropriate, part 2 (#141)

  Show description for 46b3bbd
  danobi

  authored
  46b3bbdView commit details

  Copy full SHA for 46b3bbd

  Browse repository at this point

• [allowing system role for deep seek renderer] (#139)

  ruiqi-zhong

  authored
  2e8aaa4View commit details

  Copy full SHA for 2e8aaa4

  Browse repository at this point

• Load optimizer state when appropriate (#140)

  Show description for 7728f5a
  danobi

  authored
  7728f5aView commit details

  Copy full SHA for 7728f5a

  Browse repository at this point

Commits on Nov 25, 2025

• Use stricter typing for tool calls (following kosong library) (#127)

  joschu

  authored
  53c6c38View commit details

  Copy full SHA for 53c6c38

  Browse repository at this point

Commits on Nov 23, 2025

• add support for model_path to verifiers evaluate script (#64)

  Show description for 40cf35f
  

  tmabraham

  and

  joschu

  authored
  40cf35fView commit details

  Copy full SHA for 40cf35f

  Browse repository at this point

• fix: on-policy distillation missing temperature parameter fix (#103)

  Show description for 41f93a8
  

  Andrewzh112

  and

  joschu

  authored
  41f93a8View commit details

  Copy full SHA for 41f93a8

  Browse repository at this point

• Compatible with Qwen3 multi-turn dialogue template (#102)

  gklab

  authored
  f996a59View commit details

  Copy full SHA for f996a59

  Browse repository at this point

• [tinker-cookbook] set_scope_context: add helper (#106)

  Show description for 53aaf8d
  

  3 peopleauthored
  53aaf8dView commit details

  Copy full SHA for 53aaf8d

  Browse repository at this point

• Fix rl_loop metric key to match rl/train.py convention (#125)

  Show description for 37370ca
  

  joschu

  and

  claude

  authored
  37370caView commit details

  Copy full SHA for 37370ca

  Browse repository at this point

• create extras group 'all' (#111)

  joschu

  authored
  44316a9View commit details

  Copy full SHA for 44316a9

  Browse repository at this point

• Fix TextArena double-wrapper error in TicTacToe example (#123)

  Show description for 1c0b470
  

  3 peopleauthored
  1c0b470View commit details

  Copy full SHA for 1c0b470

  Browse repository at this point

• Fix claude code OIDC issue (#124)

  joschu

  authored
  f782531View commit details

  Copy full SHA for f782531

  Browse repository at this point

• claude code: add more chat functionality, remove unnecessary perms (#122)

  joschu

  authored
  35e860dView commit details

  Copy full SHA for 35e860d

  Browse repository at this point

• [adding name to NLL evaluator] (#104)

  ruiqi-zhong

  authored
  64d4469View commit details

  Copy full SHA for 64d4469

  Browse repository at this point

• claude code review: more debugging info (#121)

  joschu

  authored
  14923caView commit details

  Copy full SHA for 14923ca

  Browse repository at this point

• claude code review: add more permissions (#120)

  joschu

  authored
  f5bb895View commit details

  Copy full SHA for f5bb895

  Browse repository at this point

• do claude code reviews when editing or reopening PR (#119)

  joschu

  authored
  6c72adbView commit details

  Copy full SHA for 6c72adb

  Browse repository at this point

• claude code auto-review, another try (#116)

  joschu

  authored
  fa9fba9View commit details

  Copy full SHA for fa9fba9

  Browse repository at this point

• hopefully improve claude code review (#115)

  joschu

  authored
  aba7733View commit details

  Copy full SHA for aba7733

  Browse repository at this point

• add proper environment for claude review (#114)

  joschu

  authored
  46f8f04View commit details

  Copy full SHA for 46f8f04

  Browse repository at this point

• auto-review with claude (#108)

  joschu

  authored
  05d3476View commit details

  Copy full SHA for 05d3476

  Browse repository at this point

• Fix verifiers-rl API compatibility for v0.1.8+ (#109)

  Show description for 5539488
  

  joschu

  and

  claude

  authored
  5539488View commit details

  Copy full SHA for 5539488

  Browse repository at this point

Commits on Nov 21, 2025

• [tinker-cookbook] supervised: add tracing annotations (#88)

  kennyyu

  authored
  5f5ce26View commit details

  Copy full SHA for 5f5ce26

  Browse repository at this point

Commits on Nov 18, 2025

• symlink CLAUDE.md to AGENTS.md (#100)

  joschu

  authored
  0adfc25View commit details

  Copy full SHA for 0adfc25

  Browse repository at this point

• qwen_instruct and search_env recipe fixes (#101)

  eddie-scio

  authored
  343f2d5View commit details

  Copy full SHA for 343f2d5

  Browse repository at this point

• Add code-rl recipe with DeepCoder (#83)

  Xiuyu-Li

  authored
  320e1c0View commit details

  Copy full SHA for 320e1c0

  Browse repository at this point

Commits on Nov 17, 2025

• Add configurable temperature parameter for RL rollout sampling (#86)

  Show description for 2ec1bcb
  

  Xiuyu-Li

  and

  joschu

  authored
  2ec1bcbView commit details

  Copy full SHA for 2ec1bcb

  Browse repository at this point

Commits on Nov 16, 2025

• Adjust sample command line (#95)

  gklab

  authored
  20e26a6View commit details

  Copy full SHA for 20e26a6

  Browse repository at this point

• Fix typo in error message: 'Unsuppoerted' -> 'Unsupported' (#94)

  BryanTegomoh

  authored
  9920715View commit details

  Copy full SHA for 9920715

  Browse repository at this point

• [Docs/Minor] Add missing tinker import (#93)

  Show description for a09cfe7
  

  vinooganesh

  and

  vinoo-ganesh-keru

  authored
  a09cfe7View commit details

  Copy full SHA for a09cfe7

  Browse repository at this point

• supervised/train.py: shift evals so they occur after 0, k, 2k, ... steps instead of 1, k+1, 2k+1, etc where k=config.eval_every (#90)

  joschu

  authored
  f624b90View commit details

  Copy full SHA for f624b90

  Browse repository at this point

Commits on Nov 14, 2025

• [Train-on-what] (#85)

  ruiqi-zhong

  authored
  6c9f7a4View commit details

  Copy full SHA for 6c9f7a4

  Browse repository at this point

Commits on Nov 13, 2025

• [math_grading] Support cross platform timeout (#92)

  Show description for e28efe2
  danobi

  authored
  e28efe2View commit details

  Copy full SHA for e28efe2

  Browse repository at this point