 Yikai Zhang
|
077fb59228
|
[AMD][LoRA] Add a gpt-oss-20b MoE LoRA RL launcher (#3157)
|
2026-09-30 21:17:08 -04:00 |
|
|
Jiajun Li
|
9e4260de04
|
Use the base image's Mooncake instead of reinstalling a dev wheel (#3738)
|
2026-09-27 21:55:14 -07:00 |
|
|
Jiajun LiandClaude Opus 5.5
|
f84b1496e1
|
test(e2e): move 8x H100 e2e tests to H200 stages and widen parallelism coverage (#3660)
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
|
2026-09-27 20:41:43 -07:00 |
|
|
fzyzcjy
|
54f7bae7c2
|
Let an infra path name the namespace it is rendered for (#3046)
|
2026-09-26 20:31:30 +08:00 |
|
|
fzyzcjy
|
5c323d6338
|
Describe a cluster's storage as a list of volumes (#3045)
|
2026-09-26 20:30:58 +08:00 |
|
|
fzyzcjy
|
821ce4af00
|
Document the cluster backend for the people who will use it (#2469)
|
2026-09-26 18:42:34 +08:00 |
|
|
Ziang Li
|
c0e607d32e
|
Limit NVFP4 conversion to main language decoder experts (#3638)
|
2026-09-24 14:56:05 -07:00 |
|
|
Yueming Yuan
|
3cdaad1027
|
Unify HF snapshot and checkpoint writing (#3198)
|
2026-09-23 18:22:35 -07:00 |
|
|
Yueming YuanandEthan Su
|
7d1d15b1cc
|
Unify LoRA target selection with HF model layouts (#3318)
Signed-off-by: Yusheng Su <yushengsu.thu@gmail.com>
Co-authored-by: Ethan (Yusheng) Su <yushengsu.thu@gmail.com>
|
2026-09-23 15:47:18 -07:00 |
|
|
Nan JiangandJiajun Li
|
9652920fc3
|
[Sampling Replay] Bounded sampling-support replay (#3354)
Co-authored-by: Jiajun Li <guapisolo@gmail.com>
|
2026-09-23 11:34:09 -07:00 |
|
|
Zhiyao JiangandXinyu Jiang
|
8d22a2c345
|
docs(AMD): Add an AMD ROCm page and fix the AMD image and hardware notes (#3209)
Co-authored-by: Xinyu Jiang <xinyuj2@andrew.cmu.edu>
|
2026-09-17 18:13:19 -07:00 |
|
|
Yueming Yuan
|
f6d0257d83
|
[multi-lora v2 7/7] Tinker gateway example and tests
|
2026-09-16 01:26:14 -07:00 |
|
|
Yueming Yuan
|
5d9f617ee2
|
[multi-lora v2 6/7] Tinker gateway runtime and entry point
|
2026-09-16 01:26:03 -07:00 |
|
|
Yueming Yuan
|
7c87754a7f
|
[multi-lora v2 4/7] Publish immutable adapter snapshots to disk
|
2026-09-16 01:25:35 -07:00 |
|
|
Yueming Yuan
|
cf787a9a01
|
[multi-lora v2 0/7] Remove the multi-LoRA v1 implementation
|
2026-09-16 01:24:46 -07:00 |
|
|
Yikai Zhang
|
664d332796
|
[AMD][LoRA] Add a Qwen3-4B LoRA RL launcher (#3183)
|
2026-09-14 14:33:01 -07:00 |
|
|
Ziang Li
|
98697a21e4
|
Restrict NVFP4 RL to routed experts (#2855)
|
2026-09-10 17:59:20 -07:00 |
|
|
fzyzcjy
|
25cf223bf0
|
Rebuild the rollout cell around an explicit five-state machine (#2098)
|
2026-09-04 14:26:48 +08:00 |
|
|
fzyzcjy
|
b53fd2cd37
|
Rename the remaining cell start and stop wording to add and dispose (#2070)
|
2026-09-04 14:17:51 +08:00 |
|
|
fzyzcjy
|
220f128185
|
Parse the raw sglang config into a dedicated pydantic model (#1993)
|
2026-09-04 13:42:49 +08:00 |
|
|
fzyzcjy
|
bf62300596
|
Delete ServerGroup and let RolloutServer own a flat list of cells (#1883)
|
2026-09-04 13:14:49 +08:00 |
|
|
fzyzcjy
|
42736b90b6
|
Wire up InferenceController and RolloutExecutor and make the controller API async (#1843)
|
2026-09-04 12:20:34 +08:00 |
|
|
Yueming Yuan
|
9f62cba24c
|
[PPO] Support the critic under --megatron-to-hf-mode bridge (#2878)
|
2026-09-02 13:43:19 -07:00 |
|
|
Ziang Liandyueming-yuan
|
75db168a27
|
Stream optimizer main initialization to NVMe (#2653)
Co-authored-by: yueming-yuan <yym022502@gmail.com>
|
2026-08-31 11:57:20 -07:00 |
|
|
Cruz ZhaoandZhichenzzz
|
128cdfb99b
|
docs: add Mooncake rollout data transfer guide (#2535)
Co-authored-by: Zhichenzzz <zczeng@uw.edu>
|
2026-08-20 15:06:29 -07:00 |
|
|
Yueming Yuan
|
748935dc71
|
docs: fix stale paths, flags, env vars and metric names (#2651)
|
2026-08-18 08:44:01 -07:00 |
|
|
Yueming Yuan
|
1beaf6cff4
|
Allow --stream-optimizer-state-to-disk without trainer offload (#2543)
|
2026-08-14 14:29:06 -07:00 |
|
|
Tao LinandClaude Fable 5
|
946ade07be
|
docs: label-only sidebar groups, explicit Overview pages, scoped link color (#2489)
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
|
2026-08-13 14:20:23 -07:00 |
|
|
Tao LinandClaude Opus 5
|
fd5ff97f36
|
docs: list Kimi-K2.6 as a P2P-supported model (#2492)
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
|
2026-08-12 17:27:34 -07:00 |
|
|
Nan JiangandZhichenzzz
|
b607ad9712
|
docs: add disaggregated RL rollout guide (#2484)
Co-authored-by: Zhichenzzz <zczeng@uw.edu>
|
2026-08-12 15:39:52 -07:00 |
|
|
Nan Jiang
|
ab9d3d8486
|
docs: rewrite INT4 QAT guide (#2478)
|
2026-08-12 13:13:56 -07:00 |
|
|
Yueming YuanandZhichen Zeng
|
2b2ef6f8b9
|
Replace all the .sh launch scripts with .py launch script (#2356)
Co-authored-by: Zhichen Zeng <zhichen.zeng@radixark.ai>
|
2026-08-11 21:51:40 -07:00 |
|
|
Ethan (Yusheng) SuandZhichen Zeng
|
0f879e5ec2
|
docs: expand LoRA training and serving guide (#2349)
Co-authored-by: Zhichen Zeng <zhichen.zeng@radixark.ai>
|
2026-08-11 20:23:00 -07:00 |
|
|
Zhichen ZengandShi Dong
|
1a13e917fe
|
docs: reorganize the Training Backends section, drop the experimental FSDP framing (#2375)
Co-authored-by: Shi Dong <shi.dong@radixark.ai>
|
2026-08-11 19:49:57 -07:00 |
|
|
Yueming Yuan
|
221dbc7eaa
|
Delete the glm4-9B, mimo-7B, moonlight-16B and deepseek-r1 launch scripts (#2354)
|
2026-08-11 18:22:26 -07:00 |
|
|
Tao LinandClaude Fable 5
|
050735a9b5
|
docs: fix SEO gaps across the docs site (#2400)
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
|
2026-08-11 17:00:26 -07:00 |
|
|
Ziang LiandZhichenzzz
|
0d95756e72
|
docs: refresh MXFP8 and NVFP4 RL guide (#2396)
Co-authored-by: Zhichenzzz <zczeng@uw.edu>
|
2026-08-11 16:54:23 -07:00 |
|
|
Yueming Yuan
|
8f37d8bec9
|
docs: replace DeepSeek V3/R1 page with a DeepSeek-V3.2 recipe (#2391)
|
2026-08-11 15:13:07 -07:00 |
|
|
Shi-Dong
|
5e14ec7cf6
|
docs: update homepage Core features section per the v0.1 feature list (#2264)
|
2026-08-10 11:55:18 -07:00 |
|
|
fzyzcjy
|
fe30593dd6
|
Replace the model config shell scripts with python (#1910)
|
2026-08-09 18:51:21 +08:00 |
|
|
maocheng23andClaude Opus 4.8
|
5400905d59
|
[OPD] Per-position teacher scoring (sparse top-k) + kaixih's robustness fixes (#1298)
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
2026-08-03 00:27:34 -07:00 |
|
|
Yueming YuanandZhichen Zeng
|
7da1079a85
|
feat(optimizer): NVMe optimizer-state streaming as a miles plugin (#1793)
Co-authored-by: Zhichen Zeng <zhichen.zeng@radixark.ai>
|
2026-07-31 22:57:57 -07:00 |
|
|
Shi-DongandZhichen Zeng
|
0916d04f2f
|
refactor(examples): group examples into infra_features/ and experimental/ (#1953)
Co-authored-by: Zhichen Zeng <zhichen.zeng@radixark.ai>
|
2026-07-29 12:10:19 -07:00 |
|
|
Zhichen Zengandyueming-yuan
|
c372e3b50c
|
feat(offload): support disk target for training-actor offload (#1575)
Co-authored-by: yueming-yuan <yym022502@gmail.com>
|
2026-07-27 15:00:36 -07:00 |
|
|
JD
|
c81e45c719
|
Add corresponding author to P2P Weight Transfer doc (#1650)
|
2026-07-13 12:52:01 -07:00 |
|
|
maocheng23
|
af28a061d0
|
[OPD] [2/N] Move to miles/rollout, add CI test and docs (#994)
|
2026-06-15 23:52:51 -07:00 |
|
|
|
04b61f31a1
|
Migrate documentation into docs/ with full history (#1321)
Co-authored-by: Zhichen Zeng <zczeng@uw.edu>
Co-authored-by: Shi Dong <shi.dong@radixark.ai>
Co-authored-by: Yueming Yuan <yym022502@gmail.com>
Co-authored-by: Shi-Dong <Shi-Dong@users.noreply.github.com>
Co-authored-by: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
Co-authored-by: Ethan (Yusheng) Su <yushengsu.thu@gmail.com>
Co-authored-by: Haoguang Cai <kai.cai@radixark.ai>
Co-authored-by: Mingyi Lu <wisclmy0611@gmail.com>
Co-authored-by: Jiajun Li <guapisolo@gmail.com>
|
2026-06-12 13:38:50 -07:00 |
|
|
|
70859d7433
|
doc [7/n]: add docs/advanced and docs/platforms (#1103)
Co-authored-by: Zhichenzzz <zhichenzzz@users.noreply.github.com>
Co-authored-by: Shi-Dong <Shi-Dong@users.noreply.github.com>
|
2026-05-11 17:46:45 -07:00 |
|