Ethan (Yusheng) Su
6efe341a05
ci(docker): bump Megatron-Bridge to bridge-new (CI probe) and testing ( #3291 )
...
Signed-off-by: Yusheng Su <yushengsu.thu@gmail.com >
2026-09-22 05:00:54 +00:00
Ethan (Yusheng) Su
ee51bfc5b3
[dockerfile] Track radixark/Megatron-Bridge@bridge in the CUDA Dockerfile ( #2872 )
2026-09-01 16:48:37 -07:00
Ethan (Yusheng) Su
31f1ca961a
[dashboard for amd gpu]: add AMD SMI GPU telemetry ( #2696 )
2026-08-21 18:09:36 -07:00
Ethan (Yusheng) Su and Jiajun Li
85e7655f13
fix(ci): calibrate lora E2E estimates from nightly runs and halve GLM5 lora matrices ( #2214 )
...
Co-authored-by: Jiajun Li <guapisolo@gmail.com >
2026-08-13 15:28:34 -07:00
Ethan (Yusheng) Su and Zhichen 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
Ethan (Yusheng) Su and Zhichen Zeng
93e71d6a6a
feat(multi-lora): enable and validate MoE expert adapters ( #1794 )
...
Co-authored-by: Zhichen Zeng <zhichen.zeng@radixark.ai >
2026-07-29 18:42:35 -07:00
Ethan (Yusheng) Su
6ccdc7da64
fix(lora): honor --attention-backend in the bridge LoRA path ( #1915 )
2026-07-28 16:30:43 -07:00
Ethan (Yusheng) Su
93a2137a0b
[fix] gate LoRA checkpoint writers on dp and cp, not dp alone ( #1722 )
2026-07-20 13:13:05 -07:00
Ethan (Yusheng) Su
49c18a2814
[fix] honor --accumulate-allreduce-grads-in-fp32 on the bridge LoRA DDP path ( #1721 )
2026-07-20 13:13:00 -07:00
Ethan (Yusheng) Su
d4adb93983
[WIP] feat: lora GLM-5.2 FP8 rollout support ( #1678 )
...
Signed-off-by: Yusheng Su <yushengsu.thu@gmail.com >
2026-07-15 16:22:36 -07:00
Ethan (Yusheng) Su
bf536263c4
[Lora] Add GDN (Qwen3.5/Qwen3-Next) LoRA target-module mapping ( #1597 )
...
Signed-off-by: Yusheng Su <yushengsu.thu@gmail.com >
2026-07-15 14:15:35 -07:00
Ethan (Yusheng) Su and Yueming Yuan
8dccb85b9f
[Lora] Fix bridge-LoRA path silently dropping recompute args ( #1593 )
...
Signed-off-by: Yusheng Su <yushengsu.thu@gmail.com >
Co-authored-by: Yueming Yuan <yym022502@gmail.com >
2026-07-07 20:11:57 -07:00
5bdd942329
Add GLM-5/5.1/5.2 (744B MoE) LoRA RL support ( #1559 )
...
Signed-off-by: Yusheng Su <yushengsu.thu@gmail.com >
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com >
Co-authored-by: yueming-yuan <yym022502@gmail.com >
2026-07-07 15:46:00 -07:00
Ethan (Yusheng) Su
44803cd0d5
Revert "[lora] avoid merge adapter weights for base weight sync" ( #1091 )
2026-05-08 07:51:28 +02:00
Ethan (Yusheng) Su
99956c06d0
[upgrade] update Megatron-Bridge source and LoRA CI to megatron e2e tests and ( #1023 )
2026-04-20 23:36:15 -07:00
Ethan (Yusheng) Su
ef5dda66b4
[Fix] fix ci ( #894 )
2026-04-05 15:46:23 -07:00
Ethan (Yusheng) Su
cb3a50ed8f
[CI] set threshold for sglang_config ci test ( #848 )
2026-03-31 22:50:56 -07:00
Ethan (Yusheng) Su
39393552cb
[feat] (1/n) fault tolerance and slime rollout structure ( #723 )
2026-03-16 23:01:53 -07:00
Ethan (Yusheng) Su
ffb61896cb
[CI] Add LoRA CI and Protection of LoRA Weight Sync ( #684 )
2026-03-06 17:42:27 -08:00
Ethan (Yusheng) Su and gongyisheng
5b59a5246a
[feat] miles lora megatron backend ( #409 )
...
Co-authored-by: gongyisheng <yishenggong9437@gmail.com >
2026-02-26 15:44:18 -08:00
Ethan (Yusheng) Su
3c56e678b3
Update Megatron-Bridge installation source in Dockerfile ( #624 )
2026-02-23 14:33:45 -08:00
Ethan (Yusheng) Su
fbb907f1b6
[bug fix] Install flashinfer_python>=0.6.2 in Dockerfile.dev ( #616 )
2026-02-18 21:03:00 -07:00
Ethan (Yusheng) Su and Yusheng Su
81ea4a82e4
[CI] Re-organize and enable necessary end2end CI cases ( #499 )
...
Co-authored-by: Yusheng Su <radixark@ac-h200-user-3.tail134ba0.ts.net >
2026-01-27 14:12:08 -08:00
Ethan (Yusheng) Su and Yusheng Su
dfd822cd94
[CI] Fix and setup CI ( #402 )
...
Co-authored-by: Yusheng Su <radixark@ac-h200-user-3.tail134ba0.ts.net >
2026-01-19 12:38:24 -08:00
Ethan (Yusheng) Su
ef57bde363
Revert "[Feat] Lora FSDP RL training - #326 and add CI/CD tests ( #351 )" ( #370 )
2025-12-28 22:54:25 -08:00
Ethan (Yusheng) Su
3da31ef394
Revert "Fix lora_rank attribute check in arguments.py" ( #369 )
2025-12-28 22:53:19 -08:00
Ethan (Yusheng) Su
54d57175a2
Revert "Add LoRA for FSDP backend. ( #307 )" [will merge on later] ( #351 )
2025-12-25 00:06:05 -08:00
22f25278e7
[Hardware] AMD - MI350/MI355 dockerfile ( #306 )
...
Co-authored-by: gramesh-amd <goramesh@amd.com >
Co-authored-by: yushengsu-thu <yushengsu.thu@gmail.com >
Co-authored-by: arist12 <ykzhang@cs.wisc.edu >
Co-authored-by: jhinpan <jpan236@wisc.edu >
Co-authored-by: zyzshishui <zyzshishui@gmail.com >
Co-authored-by: guapisolo <guapisolo@gmail.com >
Co-authored-by: sunxxuns <sunxiao04@gmail.com >
Co-authored-by: zhaochenyang20 <sunxiao04@gmail.com >
2025-12-12 12:20:10 -08:00
Ethan (Yusheng) Su
6e70fb54ac
[Hardware] AMD - pre-commit to fix the formate in commit 4f6b95f ( #302 )
2025-12-05 11:57:54 -08:00
Ethan (Yusheng) Su
5f781608ba
[Hardware] AMD Dockerfile update - support up to d4a7741 (Sep 6, 2025) ( #307 )
...
* update the dockerfile
* update the doc
* revise aiter and sglang version
* update the dockerfile
* update
* update the training script
* update docker/Dockerfile.rocm
* update
* support llama3
* update
* update
* del model_convert.sh
* add the converted model resource
* update dockerfile
* update doc
* update
* formate
* add the docker/dockerfile supportted commit ID - Need to fix commit ID:c22f55b arised issue on AMD GPU in future
* update
* update
* update the docker
* del install_megatron_core.sh
* update doc
2025-09-08 13:48:35 +08:00
Ethan (Yusheng) Su
5e41716240
[Doc typo] Update amd_tutorial.md ( #246 )
...
Fix some typo in amd_toturial
2025-09-02 17:13:04 +08:00
Ethan (Yusheng) Su
36711aa668
[Hardware] AMD Dockerfile update and test ( #158 )
...
* update the dockerfile
* update the doc
* revise aiter and sglang version
* update the dockerfile
* update
* update the training script
* update docker/Dockerfile.rocm
* update
* support llama3
* update
* update
* del model_convert.sh
* add the converted model resource
* update dockerfile
* update doc
* update
* formate
* add the docker/dockerfile supportted commit ID - Need to fix commit ID:c22f55b arised issue on AMD GPU in future
2025-08-23 09:09:55 +08:00
Ethan (Yusheng) Su
5ba5b4d1d2
[Hardware] Support AMD (ROCMm Kernel) - Update the SGLang version ( #124 )
...
* update the dockerfile
* update the doc
* revise aiter and sglang version
2025-08-01 07:44:22 +08:00
Ethan (Yusheng) Su
a5b1bd3967
[Hardware] Support AMD (ROCMm Kernel) ( #119 )
...
* update the dockerfile
* update the doc
2025-07-30 15:53:00 +08:00
Yusheng Su
2b006715ec
Merge remote-tracking branch 'upstream/main'
2025-07-02 03:32:32 +00:00
Yusheng Su
e63cdc2fad
add Acknowledgement
2025-07-02 03:30:59 +00:00
Yusheng Su
f9c67f9fa9
update
2025-07-01 08:00:05 +00:00
Yusheng Su
d885c3aecb
update dos
2025-07-01 07:52:32 +00:00
Yusheng Su
990fc633c7
recover patch files
2025-07-01 07:29:09 +00:00
Yusheng Su
a127b10405
Merge remote-tracking branch 'upstream/main'
2025-07-01 07:17:31 +00:00
Yusheng Su
c3f00b26bf
update docker
2025-07-01 07:15:46 +00:00
Yusheng Su
f803b1d51b
update readme and toturial
2025-07-01 07:12:32 +00:00
Yusheng Su
b4e669f2ec
update
2025-07-01 07:10:27 +00:00
Yusheng Su
0f0a4388d6
update amd_toturial.md
2025-07-01 01:06:48 +00:00
Yusheng Su
600336a932
support amd gpus
2025-06-30 22:54:05 +00:00
Ethan (Yusheng) Su