Commits

Branch selector

main

User selector

All users

All time

Commit history

Commits on Dec 30, 2025

• [tokenizer] update to use llama 3 instruct tokenizer by default (#260)

  Show description for 36c90b7
  

  klshuster

  and

  joschu

  authored
  36c90b7View commit details

  Copy full SHA for 36c90b7

  Browse repository at this point

Commits on Dec 29, 2025

• Add .env files to gitignore (#259)

  joschu

  authored
  207c914View commit details

  Copy full SHA for 207c914

  Browse repository at this point

• Refactor tool use recipe to use renderer's tool calling methods (#256)

  joschu

  authored
  b12c2cdView commit details

  Copy full SHA for b12c2cd

  Browse repository at this point

• Fix Qwen3 tool declaration format to match HuggingFace template (#255)

  joschu

  authored
  00758b2View commit details

  Copy full SHA for 00758b2

  Browse repository at this point

• Add progress bar to RL sampling (#257)

  joschu

  authored
  a9047a7View commit details

  Copy full SHA for a9047a7

  Browse repository at this point

• [hyperparam_utils.py] add hidden sizes for a couple models (#258)

  Show description for 5822111
  

  3 peopleauthored
  5822111View commit details

  Copy full SHA for 5822111

  Browse repository at this point

Commits on Dec 28, 2025

• Improve rl_loop.py: batch sampling, clearer naming, progress bar (#253)

  joschu

  authored
  b4c5654View commit details

  Copy full SHA for b4c5654

  Browse repository at this point

• Update docs (#252)

  joschu

  authored
  6ad2ba0View commit details

  Copy full SHA for 6ad2ba0

  Browse repository at this point

• Update changelog with renderer overhaul (#251)

  Show description for 8a13524
  

  joschu

  and

  claude

  authored
  8a13524View commit details

  Copy full SHA for 8a13524

  Browse repository at this point

• Expand renderer tests with random conversations and HF compatibility fixes (#250)

  joschu

  authored
  c6b38ceView commit details

  Copy full SHA for c6b38ce

  Browse repository at this point

Commits on Dec 27, 2025

• Improve GptOss tool calling and add Kimi K2 tool_call_id warning (#249)

  Show description for 8f918e2
  

  joschu

  and

  claude

  authored
  8f918e2View commit details

  Copy full SHA for 8f918e2

  Browse repository at this point

• Fix Kimi K2 renderer formatting and parsing (#248)

  joschu

  authored
  d63ff40View commit details

  Copy full SHA for d63ff40

  Browse repository at this point

• Add supervised/generation/parse consistency test and renderer fixes (#247)

  joschu

  authored
  00c8589View commit details

  Copy full SHA for 00c8589

  Browse repository at this point

• Rename grows_by_extension -> has_extension_property (#245)

  joschu

  authored
  6b5f74cView commit details

  Copy full SHA for 6b5f74c

  Browse repository at this point

• Add grows_by_extension property to Renderer interface (#244)

  joschu

  authored
  cc3d80aView commit details

  Copy full SHA for cc3d80a

  Browse repository at this point

• Split renderers.py into per-model-family modules (#243)

  joschu

  authored
  72b64e3View commit details

  Copy full SHA for 72b64e3

  Browse repository at this point

• shorten test_renderers.py -- split parsing tests out, remove some redundancy (#242)

  joschu

  authored
  065474eView commit details

  Copy full SHA for 065474e

  Browse repository at this point

• Add pytest CI workflow (#241)

  Show description for 7669bce
  

  joschu

  and

  claude

  authored
  7669bceView commit details

  Copy full SHA for 7669bce

  Browse repository at this point

• Refactor thinking from Message.thinking field to ThinkingPart in content list (#238)

  joschu

  authored
  0f80d40View commit details

  Copy full SHA for 0f80d40

  Browse repository at this point

Commits on Dec 26, 2025

• Fix typo in `losses.mdx` (#204)

  sunxd3

  authored
  a2b1b82View commit details

  Copy full SHA for a2b1b82

  Browse repository at this point

• Refactor render_message to use RenderContext dataclass, also fix DeepseekV3 behavior around thinking traces (#237)

  Show description for edb60a7
  

  3 peopleauthored
  edb60a7View commit details

  Copy full SHA for edb60a7

  Browse repository at this point

• Fixing broken doc link (#235)

  michaelgiba

  authored
  d3dfa43View commit details

  Copy full SHA for d3dfa43

  Browse repository at this point

• Remove automatic PR reviews, restrict @claude to trusted users (#236)

  joschu

  authored
  807e4c3View commit details

  Copy full SHA for 807e4c3

  Browse repository at this point

• Rename RenderedMessage fields: prefix/content/suffix → header/output/stop_overlap (#220)

  Show description for 379e696
  

  joschu

  and

  claude

  authored
  379e696View commit details

  Copy full SHA for 379e696

  Browse repository at this point

• Add tool specification helpers and comprehensive tool calling support (#221)

  joschu

  authored
  a23f9dcView commit details

  Copy full SHA for a23f9dc

  Browse repository at this point

Commits on Dec 25, 2025

• [logtree][table_from_dict should respect `_is_logging_enabled`] (#228)

  ruiqi-zhong

  authored
  51d4f06View commit details

  Copy full SHA for 51d4f06

  Browse repository at this point

• [utils.logtree][flushign results] (#225)

  ruiqi-zhong

  authored
  d2adfe5View commit details

  Copy full SHA for d2adfe5

  Browse repository at this point

Commits on Dec 24, 2025

• Qwen3 tool-calling fix (args -> arguments) (#222)

  tyler-griggs

  authored
  5ab4301View commit details

  Copy full SHA for 5ab4301

  Browse repository at this point

• Fix tool calling bugs (#219)

  Show description for 02f4620
  

  joschu

  and

  claude

  authored
  02f4620View commit details

  Copy full SHA for 02f4620

  Browse repository at this point

• Add learning rate config support for Moonshot Kimi models (#200)

  bledden

  authored
  bf027efView commit details

  Copy full SHA for bf027ef

  Browse repository at this point

• Fix plural form in docs-outline.mdx (#212)

  Show description for 75b11be
  

  claudeaceae

  and

  claude

  authored
  75b11beView commit details

  Copy full SHA for 75b11be

  Browse repository at this point

• Fix subject-verb agreement in math_rl README (#211)

  Show description for 0575d4c
  

  claudeaceae

  and

  claude

  authored
  0575d4cView commit details

  Copy full SHA for 0575d4c

  Browse repository at this point

• Remove duplicate word in model-lineup.mdx (#214)

  Show description for 27a6993
  

  claudeaceae

  and

  claude

  authored
  27a6993View commit details

  Copy full SHA for 27a6993

  Browse repository at this point

• Improve phrasing in tool_use/search README (#213)

  Show description for e3e3746
  

  claudeaceae

  and

  claude

  authored
  e3e3746View commit details

  Copy full SHA for e3e3746

  Browse repository at this point

• Remove duplicate word in llms-full.txt (#215)

  Show description for c6aaada
  

  claudeaceae

  and

  claude

  authored
  c6aaadaView commit details

  Copy full SHA for c6aaada

  Browse repository at this point