Commits

Breadcrumbs

History for

onmain

User selector

All users

All time

Commit history

Commits on May 11, 2026

• Fix `get_lora_param_count` with model-specific per-component table (#695)

  Show description for 76133f8
  

  nealwu

  and

  claude

  authored
  76133f8View commit details

  Copy full SHA for 76133f8

  View code at this pointBrowse repository at this point

Commits on May 2, 2026

• Fix failing CI tests (#691)

  Show description for 7bebaf6
  

  nealwu

  and

  claude

  authored
  7bebaf6View commit details

  Copy full SHA for 7bebaf6

  View code at this pointBrowse repository at this point

Commits on Apr 20, 2026

• feat: add Kimi-K2.6 model support (#675)

  Show description for 38ff1db
  

  YujiaBao

  and

  claude

  authored
  38ff1dbView commit details

  Copy full SHA for 38ff1db

  View code at this pointBrowse repository at this point

Commits on Apr 15, 2026

• feat: add Nemotron-3 low thinking renderer for Super model (#667)

  Show description for 0dd21a4
  

  YujiaBao

  and

  claude

  authored
  0dd21a4View commit details

  Copy full SHA for 0dd21a4

  View code at this pointBrowse repository at this point

Commits on Apr 12, 2026

• fix: normalize SFT weights per-example for consistent gradient magnitudes (#334)

  Show description for 23b9fd8
  

  bledden

  and

  claude

  authored
  23b9fd8View commit details

  Copy full SHA for 23b9fd8

  View code at this pointBrowse repository at this point

Commits on Apr 10, 2026

• Consolidate 7 skills into 2 (research + debug) and fix plugin manifest (#651)

  Show description for 4a93d17
  

  YujiaBao

  and

  claude

  authored
  4a93d17View commit details

  Copy full SHA for 4a93d17

  View code at this pointBrowse repository at this point

Commits on Apr 8, 2026

• Improve tic-tac-toe self-play recipe (#634)

  Show description for bd66e9e
  

  YujiaBao

  and

  claude

  authored
  bd66e9eView commit details

  Copy full SHA for bd66e9e

  View code at this pointBrowse repository at this point

• Add public API exports to distillation module and eval exceptions (#639)

  Show description for 4198f94
  

  YujiaBao

  and

  claude

  authored
  4198f94View commit details

  Copy full SHA for 4198f94

  View code at this pointBrowse repository at this point

• Split math_env_test into unit tests and integration test; consolidate CI workflows (#640)

  Show description for eb0bb2f
  

  YujiaBao

  and

  claude

  authored
  eb0bb2fView commit details

  Copy full SHA for eb0bb2f

  View code at this pointBrowse repository at this point

Commits on Apr 4, 2026

• Fix rolling checkpoint test TTL to meet API minimum (#603)

  Show description for 872a337
  

  YujiaBao

  and

  claude

  authored
  872a337View commit details

  Copy full SHA for 872a337

  View code at this pointBrowse repository at this point

Commits on Apr 3, 2026

• Fix FP8 expert quantization for fused 3D expert tensors (#597)

  Show description for c90ef8e
  

  YujiaBao

  and

  claude

  authored
  c90ef8eView commit details

  Copy full SHA for c90ef8e

  View code at this pointBrowse repository at this point

• Add structured GPU e2e tests for weight export (#585)

  Show description for 605e94c
  

  YujiaBao

  and

  claude

  authored
  605e94cView commit details

  Copy full SHA for 605e94c

  View code at this pointBrowse repository at this point

Commits on Apr 2, 2026

• Pluggable quantization formats + GPU acceleration for weights module (#583)

  Show description for cc7b6b1
  

  YujiaBao

  and

  claude

  authored
  cc7b6b1View commit details

  Copy full SHA for cc7b6b1

  View code at this pointBrowse repository at this point

• Fix tokenizer_class corrupted to TokenizersBackend during export (#582)

  Show description for 0056092
  

  YujiaBao

  and

  claude

  authored
  0056092View commit details

  Copy full SHA for 0056092

  View code at this pointBrowse repository at this point

• Fix multi-teacher distillation smoke test timeout (#581)

  Show description for de822a6
  

  YujiaBao

  and

  claude

  authored
  de822a6View commit details

  Copy full SHA for de822a6

  View code at this pointBrowse repository at this point

• Fix experts-fp8 compression_config for non-DeepSeek models (#580)

  Show description for 3fc17c7
  

  YujiaBao

  and

  claude

  authored
  3fc17c7View commit details

  Copy full SHA for 3fc17c7

  View code at this pointBrowse repository at this point

• Add Kimi K2 / K2.5 shard-by-shard merge with INT4 expert dequant/requant (#573)

  Show description for b6b3e99
  

  YujiaBao

  and

  claude

  authored
  b6b3e99View commit details

  Copy full SHA for b6b3e99

  View code at this pointBrowse repository at this point

Commits on Mar 30, 2026

• Shorten skill names for better slash command UX (#568)

  Show description for 7cb5be2
  

  YujiaBao

  and

  claude

  authored
  7cb5be2View commit details

  Copy full SHA for 7cb5be2

  View code at this pointBrowse repository at this point

• Make image_processor optional for qwen3_vl renderers (#566)

  Show description for 0817a03
  

  YujiaBao

  and

  claude

  authored
  0817a03View commit details

  Copy full SHA for 0817a03

  View code at this pointBrowse repository at this point

Commits on Mar 27, 2026

• Add global install for Claude Code skills (#545)

  Show description for 5f4be66
  

  YujiaBao

  and

  claude

  authored
  5f4be66View commit details

  Copy full SHA for 5f4be66

  View code at this pointBrowse repository at this point

• Align synthetic expert adapters with real Tinker broadcast patterns (#556)

  Show description for 3d6c026
  

  YujiaBao

  and

  claude

  authored
  3d6c026View commit details

  Copy full SHA for 3d6c026

  View code at this pointBrowse repository at this point

• Add rolling checkpoints for cheaper training resume (#534)

  Show description for 9be549f
  

  YujiaBao

  and

  claude

  authored
  9be549fView commit details

  Copy full SHA for 9be549f

  View code at this pointBrowse repository at this point

• Fix build_hf_model for Qwen3.5 MoE and add partial LoRA coverage tests (#554)

  Show description for 78b0202
  

  YujiaBao

  and

  claude

  authored
  78b0202View commit details

  Copy full SHA for 78b0202

  View code at this pointBrowse repository at this point

• Add e2e test for adapter upload to HuggingFace Hub (#544)

  Show description for ce035ba
  

  YujiaBao

  and

  claude

  authored
  ce035baView commit details

  Copy full SHA for ce035ba

  View code at this pointBrowse repository at this point

• Add model card generation for HuggingFace Hub publishing and e2e tests (#543)

  Show description for c9344b1
  

  YujiaBao

  and

  claude

  authored
  c9344b1View commit details

  Copy full SHA for c9344b1

  View code at this pointBrowse repository at this point

• Fix build_hf_model for Nemotron fused Mamba projections and backbone prefix (#549)

  Show description for 4e68f36
  

  YujiaBao

  and

  claude

  authored
  4e68f36View commit details

  Copy full SHA for 4e68f36

  View code at this pointBrowse repository at this point

• Fix build_lora_adapter for Nemotron empty expert LoRA and fused Mamba projections (#548)

  Show description for 125e52e
  

  YujiaBao

  and

  claude

  authored
  125e52eView commit details

  Copy full SHA for 125e52e

  View code at this pointBrowse repository at this point

Commits on Mar 25, 2026

• Add Nemotron-3-Super vLLM test and update model docs (#542)

  Show description for 67faf93
  

  YujiaBao

  and

  claude

  authored
  67faf93View commit details

  Copy full SHA for 67faf93

  View code at this pointBrowse repository at this point

• Add Nemotron-3 adapter serving, vLLM e2e tests, and weights README (#539)

  Show description for f6bf8d2
  

  YujiaBao

  and

  claude

  authored
  f6bf8d2View commit details

  Copy full SHA for f6bf8d2

  View code at this pointBrowse repository at this point

• Add test_lifecycle to weights-e2e workflow and pass TINKER_API_KEY (#536)

  Show description for f97d078
  

  YujiaBao

  and

  claude

  authored
  f97d078View commit details

  Copy full SHA for f97d078

  View code at this pointBrowse repository at this point

Commits on Mar 24, 2026

• Introduce ActionExtra TypedDict for Env.step extensibility (#538)

  Show description for a33bb77
  

  YujiaBao

  and

  claude

  authored
  a33bb77View commit details

  Copy full SHA for a33bb77

  View code at this pointBrowse repository at this point

• Remove unused `name` parameter from `save_weights_and_get_sampling_client` calls (#535)

  Show description for de0a7b6
  

  YujiaBao

  and

  claude

  authored
  de0a7b6View commit details

  Copy full SHA for de0a7b6

  View code at this pointBrowse repository at this point

• Implement build_lora_adapter for PEFT-format adapter serving (#533)

  Show description for a87a9f5
  

  YujiaBao

  and

  claude

  authored
  a87a9f5View commit details

  Copy full SHA for a87a9f5

  View code at this pointBrowse repository at this point

• Handle context limits and max-tokens truncation in multi-turn RL environments (#506)

  Show description for c53f137
  

  YujiaBao

  and

  claude

  authored
  c53f137View commit details

  Copy full SHA for c53f137

  View code at this pointBrowse repository at this point

• Split weights e2e tests by model family and add CI workflow (#531)

  Show description for 3b020ad
  

  YujiaBao

  and

  claude

  authored
  3b020adView commit details

  Copy full SHA for 3b020ad

  View code at this pointBrowse repository at this point