Skip to content

Pull requests: ROCm/ATOM

New pull request New
82 Open 509 Closed
82 Open 509 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

feat(deepseek_v4): FP8 batched BMM for grouped output LoRA wo_a
#676 opened May 1, 2026 by zufayu Loading…
2
Enable Cohere Command-R (CohereForCausalLM / Cohere2ForCausalLM) on ATOM
#675 opened Apr 30, 2026 by jatseng-ai Loading…
support swiglu a4w4 moe
#674 opened Apr 30, 2026 by XiaobingSuper Contributor • Draft
1 task
[atom-vllm benchmark][docker release] refine docker release and atom-vllm benchmark details
#671 opened Apr 29, 2026 by zejunchen-zejun Collaborator • Draft
14
[fix][acc] fix accuracy of fp8 attn weights model using ptpc quant recipe
#670 opened Apr 29, 2026 by gbyu-amd Contributor Loading…
1 task
[vllm-atom] Fix GLM-5 accuracy in vLLM plugin
#669 opened Apr 29, 2026 by kliuae-amd Contributor Loading…
1 task
dsa: remove block_table_convert_triton in dsa
#658 opened Apr 28, 2026 by junhaha666 Contributor Loading…
1 task
9
prefill gdr kernel enablement
#656 opened Apr 28, 2026 by ganyi1996ppo Contributor Loading…
1 task
2
Support Mimo-v2.5-Pro
#654 opened Apr 28, 2026 by wufann Contributor • Draft
1 task
[Enhancement] support online quantization
#653 opened Apr 28, 2026 by haoyangli0109 Contributor • Draft
feat(deepseek_v4): PR1 skeleton — end-to-end inference with triton MoE
#650 opened Apr 25, 2026 by valarLip Collaborator • Draft
5 of 7 tasks
2
support sparse attn mtp
#649 opened Apr 25, 2026 by jiayyu Contributor Loading…
1 task
8
[vLLM-ATOM] Add profile trace parsing tool for vLLM-ATOM
#644 opened Apr 24, 2026 by kliuae-amd Contributor • Draft
1 task
[Draft][ATOM-SGLang][Feat] Enable Deepseek v3 MTP
#643 opened Apr 24, 2026 by ZhiweiYan-96 Contributor • Draft
4
feat: add Step-3.5-Flash support and fix MoE weight shuffling on gfx950
#641 opened Apr 23, 2026 by LJ-underdog • Draft
1 task
4
[benchmark] Add TP8 EP8 case for deepseek FP4
#639 opened Apr 23, 2026 by qichu-yun Contributor Loading…
1 task
fix: correct MXFP4 MoE weight shuffle for Quark models (MiniMax-M2.1)
#632 opened Apr 22, 2026 by thpereir Contributor • Draft
1 task
[Kimi] support Eagle3 speculative decoding for Kimi K2.5
#631 opened Apr 22, 2026 by yhl-amd Loading…
1
Gemma16w16 integration
#627 opened Apr 21, 2026 by amirumoAMD Contributor • Draft
1 task
[atom-vllm] use aiter greedy sampler
#622 opened Apr 21, 2026 by zejunchen-zejun Collaborator • Draft
(ci)(recipe): Add DeepSeek-R1 FP4 TP4 validation and DS recipe for SGLang-ATOM
#614 opened Apr 20, 2026 by zhuyuhua-v Collaborator • Draft
12
[feat](minimax): refactor rmsnorm for minimax
#613 opened Apr 20, 2026 by PerryZhang01 Contributor Loading…
2
mla: drop max_split_per_batch=16 cap to match vLLM
#611 opened Apr 20, 2026 by peizhang56 Loading…
3 tasks
1
[feat](ai): add accuracy debug skill for nightly test
#607 opened Apr 19, 2026 by PerryZhang01 Contributor Loading…
[plugin] Flux2 model support
#606 opened Apr 19, 2026 by Phi-C • Draft
ProTip! Follow long discussions with comments:>50.