19 Commits
Author SHA1 Message Date
Xinyu JiangandZhiyao Jiang 874b3e2756 [AMD] docker: build the ROCm images on the sglang v0.5.20 release images (#3873)
Co-authored-by: Zhiyao Jiang <jessicajiang324@gmail.com>
2026-09-30 00:10:03 -07:00
Xinyu JiangandZhiyao Jiang 5ebc06f8d2 [AMD] ci: expand ROCm nightly coverage and refresh runtime estimates (#2541)
Co-authored-by: Zhiyao Jiang <jessicajiang324@gmail.com>
2026-09-29 17:28:58 -07:00
Xinyu JiangandZhiyao Jiang e30b8b6dd6 [AMD] replace the ROCm 7.2.0 image with ROCm 7.2.4 (#3326)
Co-authored-by: Zhiyao Jiang <jessicajiang324@gmail.com>
2026-09-24 03:59:24 -04:00
Xinyu JiangandZhiyao Jiang 8d9826eacc [AMD] docker: retire the rocm700 variants, add a rocm10-mi35x variant (#2854)
Co-authored-by: Zhiyao Jiang <jessicajiang324@gmail.com>
2026-09-08 22:37:39 -07:00
Xinyu JiangandZhiyao Jiang babb429038 [AMD CI] Enable verified ROCm tests (#2480)
Co-authored-by: Zhiyao Jiang <jessicajiang324@gmail.com>
2026-08-14 13:05:50 -07:00
Xinyu JiangandZhiyao Jiang d9e3c3b394 [AMD] DeepSeek-V4: use the ROCm precision-parity norm path (#2040)
Co-authored-by: Zhiyao Jiang <jessicajiang324@gmail.com>
2026-07-31 15:09:50 -07:00
Xinyu Jiang c158722a05 [AMD] fix mooncake object store support (#2039) 2026-07-31 15:09:41 -07:00
Xinyu JiangandZhiyao Jiang 310ec07c0a Fix weight sync selector for frozen speculative drafts (#1926)
Co-authored-by: Zhiyao Jiang <jessicajiang324@gmail.com>
2026-07-29 17:34:54 -07:00
6bbc317fc1 [AMD] Drop inert DSv4 rollout knobs and add an MTP recipe (#1733)
Co-authored-by: Zhiyao Jiang <jessicajiang324@gmail.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-29 13:14:58 -07:00
Xinyu JiangandZhiyao Jiang ef1bf5f7c2 [AMD] rocm720 MI355X image: Transformer Engine wheel + build fixes (#1713)
Co-authored-by: Zhiyao Jiang <jessicajiang324@gmail.com>
2026-07-17 18:25:35 -07:00
Xinyu JiangandZhiyao Jiang 7e575f0549 [AMD] use prebuilt wheels for flash-attn and sgl-model-gateway (#1686)
Co-authored-by: Zhiyao Jiang <jessicajiang324@gmail.com>
2026-07-15 16:59:08 -07:00
Xinyu JiangandZhiyao Jiang 50a3733e2b [AMD] fix GPU memory release for TMS pause on ROCm 7.2 (#1680)
Co-authored-by: Zhiyao Jiang <jessicajiang324@gmail.com>
2026-07-15 16:57:15 -07:00
Xinyu JiangandZhiyao Jiang 352b4aafcb [AMD] Enable DeepSeek-V4-Flash FP8 RL training on MI355X (#1607)
Co-authored-by: Zhiyao Jiang <jessicajiang324@gmail.com>
2026-07-09 19:49:03 -07:00
Xinyu JiangandZhiyao Jiang 39a1580e2c [AMD] Use bridge branch for ROCm Megatron-Bridge (#1475)
Co-authored-by: Zhiyao Jiang <jessicajiang324@gmail.com>
2026-06-25 21:41:04 -07:00
Xinyu JiangandZhiyao Jiang 03fe6ae9f8 [AMD CI] [2/N] Add ROCm CI support for verified MI350X e2e tests (#1379)
Co-authored-by: Zhiyao Jiang <jessicajiang324@gmail.com>
2026-06-23 19:32:26 -07:00
Xinyu JiangandZhiyao Jiang 402a72d8d0 [AMD CI] [1/N] Support Python 3.10 on ROCm base (#1339)
Co-authored-by: Zhiyao Jiang <jessicajiang324@gmail.com>
2026-06-17 19:44:56 -07:00
Xinyu JiangandZhiyao Jiang 317bc3345d rocm: honor HIP_VISIBLE_DEVICES in ray gpu helpers and NOSET list (#1233)
Co-authored-by: Zhiyao Jiang <jessicajiang324@gmail.com>
2026-06-03 19:06:41 -07:00
Xinyu Jiang 839f32a30c [lint] black-format run_simple_amd_triton.py (#1051) 2026-04-27 17:02:48 -07:00
d74d25a620 [AMD] Support AMD True on-policy Using Triton Backend (Move into Docker patch) (#1047)
Co-authored-by: Zhiyao Jiang <jessicajiang324@gmail.com>
Co-authored-by: Yusheng Su <yushengsu.thu@gmail.com>
2026-04-26 22:44:32 -07:00