Skip to content

Pull requests: Dao-AILab/quack

New pull request New
12 Open 89 Closed
12 Open 89 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

[Rotary] Fuse copy into kernel
#144 opened May 24, 2026 by simveit Contributor Loading…
[WIP] Add non-gated colvec_scale support to gemm_dact
#143 opened May 19, 2026 by JungHoyoun • Draft
[WIP] Blockscaled SM90
#141 opened May 17, 2026 by KareemMusleh Contributor Loading…
10
Add LayerNorm Backward
#136 opened May 9, 2026 by ighoshsubho Loading…
Fix the SwiGLU/dSwiGLU bug for the (2, 2) epilogue layout
#133 opened May 7, 2026 by GarlGuo Member Loading…
1
fix the SM100 2CTA issue for SwiGLU/dSwiGLU
#131 opened May 6, 2026 by GarlGuo Member Loading…
3
SM120 blockscaled: add opt-in packed-LDSM performance path (2.3 to 103.4 TFLOP/s)
#128 opened May 1, 2026 by alecco Contributor • Draft
Add Sm120 blockscaled FP4 GEMM path
#127 opened Apr 30, 2026 by alecco Contributor • Draft
1
Refactor GEMM for flexible work, operand routing, and epilogue policies
#111 opened Apr 19, 2026 by santoshmo Contributor Loading…
Rmsnorm backward fusing sum
#101 opened Apr 8, 2026 by AaronWang04 Loading…
12
fix bug in online_softmax when all loaded values in a warp are -inf
#64 opened Jan 8, 2026 by LitLeo Loading…
6
Fused Add + RMSNorm pattern
#55 opened Nov 9, 2025 by AndreSlavescu Loading…
3
ProTip! Follow long discussions with comments:>50.