Commits

Branch selector

main

User selector

All users

All time

Commit history

Commits on Apr 1, 2026

• Multi-Agent Train, Phase 1

  

  xinglu

  committed
  d386527View commit details

  Copy full SHA for d386527

  Browse repository at this point

• feat(infra): add client-side fetch buffer for RTensor (#1122)

  Show description for 44d54cf
  guozhihao-224

  authored
  44d54cfView commit details

  Copy full SHA for 44d54cf

  Browse repository at this point

• refactor(agents): redesign review-pr taxonomy and sync flow (#1124)

  Show description for 8357f71
  

  Wentai Zhang

  authored
  8357f71View commit details

  Copy full SHA for 8357f71

  Browse repository at this point

• refactor(infra): decompose rpc_server into shared guard + blueprints (#1126)

  Show description for 8d6e4fd
  garrett4wade

  authored
  8d6e4fdView commit details

  Copy full SHA for 8d6e4fd

  Browse repository at this point

• feat(utils): add Trackio experiment tracking backend (#1113)

  Show description for c4ccc39
  guozhihao-224

  authored
  c4ccc39View commit details

  Copy full SHA for c4ccc39

  Browse repository at this point

• feat(inference_service): add VLM image input support to OpenAI-compatible API (#1119)

  Show description for 92c467e
  garrett4wade

  authored
  92c467eView commit details

  Copy full SHA for 92c467e

  Browse repository at this point

Commits on Mar 31, 2026

• fix(archon): harden FP8 blockwise training for TP and MoE scenarios (#1118)

  Show description for 0ee8562
  

  Wentai Zhang

  authored
  0ee8562View commit details

  Copy full SHA for 0ee8562

  Browse repository at this point

• feat(inference-service): complete vLLM backend support in inference service (#1112)

  Show description for d366571
  garrett4wade

  authored
  d366571View commit details

  Copy full SHA for d366571

  Browse repository at this point

• chore(ci): update GCP CI image (#1115)

  garrett4wade

  authored
  59f4047View commit details

  Copy full SHA for 59f4047

  Browse repository at this point

• feat(archon): add FP8 blockwise training support (#1087)

  Show description for f6331e0
  

  Wentai Zhang

  authored
  f6331e0View commit details

  Copy full SHA for f6331e0

  Browse repository at this point

• fix(engine): restore SGLang VLM training (#1098)

  Show description for a043177
  garrett4wade

  authored
  a043177View commit details

  Copy full SHA for a043177

  Browse repository at this point

Commits on Mar 30, 2026

• feat(fsdp): pipeline weight sync with single pending bucket (#1074)

  Show description for 2ddd959
  HT-Yuan

  authored
  2ddd959View commit details

  Copy full SHA for 2ddd959

  Browse repository at this point

• refactor(engine): abstract CUDA calls via current_platform in PerLayerOptimWrapper (#1108)

  Show description for 61281ba
  guozhihao-224

  authored
  61281baView commit details

  Copy full SHA for 61281ba

  Browse repository at this point

• fix(archon): add missing POST /data/batch endpoint to data proxy (#1105)

  Show description for cbe35f5
  

  Wentai Zhang

  authored
  cbe35f5View commit details

  Copy full SHA for cbe35f5

  Browse repository at this point

• fix(dataloader): prevent data drop and padding during validation for accurate metrics(pre-commit checked) (#1100)

  Anguo-star

  authored
  0405b5cView commit details

  Copy full SHA for 0405b5c

  Browse repository at this point

• fix(engine): remove duplicate trust_remote_code kwarg in MegatronBridge init (#1107)

  

  Wentai Zhang

  authored
  fdca82dView commit details

  Copy full SHA for fdca82d

  Browse repository at this point

• feat: megatron bridge adaptation (#1056)

  Show description for 722e235
  

  gursimar

  and

  garrett4wade

  authored
  722e235View commit details

  Copy full SHA for 722e235

  Browse repository at this point

• fix: fix pad_packed_tensor_dict (#1104)

  Show description for be621da
  

  HKAB

  and

  truongnp5

  authored
  be621daView commit details

  Copy full SHA for be621da

  Browse repository at this point

• feat: support model training in IPv6-only environment (#1072)

  Show description for c961323
  

  3 peopleauthored
  c961323View commit details

  Copy full SHA for c961323

  Browse repository at this point

Commits on Mar 27, 2026

• feat: add BailingMoeV2.5 support with Lightning Attention + MLA + MoE + CP (#1079)

  Show description for 483a4e8
  

  3 peopleauthored
  483a4e8View commit details

  Copy full SHA for 483a4e8

  Browse repository at this point

• refactor: use pause_generation instead of abort_all_req in areal_vllm_server (#1091)

  HwVanICI

  authored
  6113c02View commit details

  Copy full SHA for 6113c02

  Browse repository at this point

• fix(fsdp): stabilize qwen-vl get_rope_index args and dtype (#1094)

  Adiactive

  authored
  45e805bView commit details

  Copy full SHA for 45e805b

  Browse repository at this point

Commits on Mar 26, 2026

• fix(rpc): batch HTTP RTensor fetches for multi-image payloads (#1077)

  Wangxiaoxiaoa

  authored
  f6556e8View commit details

  Copy full SHA for f6556e8

  Browse repository at this point

• feat(commands): add fork workflow support to create-pr skill (#1092)

  Show description for 3142b88
  guozhihao-224

  authored
  3142b88View commit details

  Copy full SHA for 3142b88

  Browse repository at this point

Commits on Mar 25, 2026

• feat(platform): add NUMA CPU affinity binding for training engines (#1083)

  HT-Yuan

  authored
  aebe13dView commit details

  Copy full SHA for aebe13d

  Browse repository at this point

Commits on Mar 24, 2026

• chore(agents): add Codex harness and align AI workflows (#1082)

  Show description for 682d564
  

  Wentai Zhang

  authored
  682d564View commit details

  Copy full SHA for 682d564

  Browse repository at this point

• refactor(api): migrate allocation_mode to per-engine backend fields (#1044)

  Show description for 93b572d
  garrett4wade

  authored
  93b572dView commit details

  Copy full SHA for 93b572d

  Browse repository at this point

• fix: add PIL image and processor serialization for VLM RPC (#1070)

  Show description for 0a54cfa
  Adiactive

  authored
  0a54cfaView commit details

  Copy full SHA for 0a54cfa

  Browse repository at this point

• fix(openai): handle streaming responses in chat/completions endpoint (#1053)

  Show description for 99040b9
  Zijun9

  authored
  99040b9View commit details

  Copy full SHA for 99040b9

  Browse repository at this point

• docs: add gitcgr code graph badge (#1073)

  vitali87

  authored
  421e4e5View commit details

  Copy full SHA for 421e4e5

  Browse repository at this point

• fix(rpc): resolve connection reset during RTensor fetch with large payloads (#1075)

  Show description for f293ceb
  

  pratyush618

  and

  gemini-code-assist[bot]

  authored
  f293cebView commit details

  Copy full SHA for f293ceb

  Browse repository at this point

Commits on Mar 23, 2026

• fix(archon): simplify RTensor serialization in data proxy (#1067)

  Show description for 9639749
  garrett4wade

  authored
  9639749View commit details

  Copy full SHA for 9639749

  Browse repository at this point

Commits on Mar 20, 2026

• test(infra): speed up inference service integration tests (#1068)

  Show description for 3bf10c9
  

  nuzant

  authored
  3bf10c9View commit details

  Copy full SHA for 3bf10c9

  Browse repository at this point

Commits on Mar 19, 2026

• feat: estimators for kl divergence (#1060)

  Show description for dfeab63
  

  NicolasArias

  and

  RedDreamer

  authored
  dfeab63View commit details

  Copy full SHA for dfeab63

  Browse repository at this point

• feat(gateway): Add rollout gateway infrastructure with controller, router, and data proxy (#1043)

  Show description for 4c9ec37
  

  nuzant

  authored
  4c9ec37View commit details

  Copy full SHA for 4c9ec37

  Browse repository at this point