Skip to content

Pull requests: mirage-project/mirage

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
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort

Pull requests list

Raise the SM100 attention max_num_batched_tokens ceiling
#778 opened Sep 19, 2026 by JerryLiang49 Contributor Loading…
Fix Hopper multitoken_paged_attention mbt>1 stale-tail NaN
#768 opened Sep 14, 2026 by jyizheng Contributor Loading…
Full unified KV pool support and migration
#767 opened Sep 13, 2026 by xiaozr Collaborator Loading…
Lowering to Superoptimizer IR
#759 opened Sep 1, 2026 by aramesh10 Collaborator Draft
test(serving): cover online request overlap in CI
#756 opened Aug 26, 2026 by AravTewari Collaborator Loading…
Unified KV Cache Pool for Hybrid KV Specs
#747 opened Jul 27, 2026 by xiaozr Collaborator Loading…
Qwen3.5 support by mpk skill set
#746 opened Jul 27, 2026 by bill810975 Contributor Draft
Deepseek v3 support
#743 opened Jul 20, 2026 by bill810975 Contributor Loading…
Add SM120 support
#738 opened Jul 7, 2026 by Snowfall99 Loading…
Adding Structured Decoding with XGrammar
#736 opened Jun 30, 2026 by aramesh10 Collaborator Loading…
SM100 Linear Layer in CUDA
#729 opened Jun 18, 2026 by aramesh10 Collaborator Loading…
Revert/dpskv3
#718 opened Jun 11, 2026 by ZephyrZhao247 Collaborator Loading…
Feat/qwen3 new api
#717 opened Jun 11, 2026 by ZephyrZhao247 Collaborator Draft
Feat/new api
#708 opened Jun 5, 2026 by ZephyrZhao247 Collaborator Draft
NVFP4 + MXFP4 Linear Layer
#694 opened May 25, 2026 by aramesh10 Collaborator Loading…
Add SM100 planned attention support
#693 opened May 21, 2026 by tangzhanshuo Loading…
build: rework wheel release pipeline and clean up CI helpers
#680 opened May 4, 2026 by Marco-Christiani Contributor Loading…
mHC Support for DeepSeek v4
#677 opened May 4, 2026 by aramesh10 Collaborator Loading…
Hash Routing Support for DeepSeek V4 Flash
#675 opened May 3, 2026 by STWMichael Collaborator Draft
TPU Transpiler
#654 opened Apr 13, 2026 by STWMichael Collaborator Draft
Expert parallelism: Dispatch and Combine
#653 opened Apr 12, 2026 by ZephyrZhao247 Collaborator Draft
[Blackwell][SM100] FP8 quantize and dense GEMM runtime kernels
#647 opened Mar 31, 2026 by JudySun233 Collaborator Loading…
EP MoE
#631 opened Feb 25, 2026 by STWMichael Collaborator Draft
swizzling attn tasks to improve prefill performance
#608 opened Jan 17, 2026 by wenqiny Contributor Draft
ProTip! What’s not been updated in a month: updated:<2026-08-22.