Skip to content

Pull requests: flashinfer-ai/flashinfer

New pull request New
78 Open 1,706 Closed
78 Open 1,706 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

Fix autotuner oom
#2442 opened Jan 30, 2026 by zack041 Loading…
3 of 5 tasks
1
4
Add sm90 guard to fence ptx run-ci
#2439 opened Jan 29, 2026 by jhalabi-nv Loading…
5 tasks
9
fix: improve numerical stability of Gumbel sampling run-ci
#2438 opened Jan 29, 2026 by ixlmar Loading…
4 of 5 tasks
12
fix: Sampling: CUDA Graph fix
#2432 opened Jan 29, 2026 by IzzyPutterman Loading…
5 tasks
9
refactor: refactoring cuda code to cute-dsl (part 1)
#2428 opened Jan 28, 2026 by yzh119 Loading…
5 tasks
33
[wip] Topk weights
#2425 opened Jan 27, 2026 by aleozlx • Draft
5 tasks
2
refactor: reduce hopper's gdn prefill compilation time and fix docstring.
#2422 opened Jan 27, 2026 by yzh119 Loading…
5 tasks
13
infra: add manual code owner override support in codeowner_analyzer.py
#2418 opened Jan 26, 2026 by sricketts Loading…
3 tasks done
23
[CI]: Enable Blackwell & Hopper in public CI testing
#2413 opened Jan 24, 2026 by yongwww Loading…
5 tasks
5
chore: update benchmark scripts; fix trtllm-gen moe comments
#2412 opened Jan 24, 2026 by IwakuraRein Loading…
5 tasks done
14
Add/update multi node/multi GPU test scripts
#2410 opened Jan 23, 2026 by dierksen Loading…
3 of 5 tasks
11
perf: improve gdn decode cute-dsl kernels
#2405 opened Jan 23, 2026 by yzh119 Loading…
4 of 5 tasks
20
feat: cuteDSL fp4 moe for better DSR1 performance.
#2398 opened Jan 22, 2026 by nv-yunzheq Loading…
5 tasks
64
unify memory allocation for single-node and multi-node nvlink all reduce(+fusion) using torch symmetric memory
#2389 opened Jan 20, 2026 by Amir-19 • Draft
5 tasks
7
feat: add trtllm_fp8_block_scale_routed_moe API
#2382 opened Jan 20, 2026 by yzh119 Loading…
1
9
Improve sampling benchmarks.
#2374 opened Jan 19, 2026 by vincentzed • Draft
5 tasks
7
NVFP4 KV Cache on SM100
#2363 opened Jan 16, 2026 by samuellees • Draft
5 tasks
7
feat: add per-request generator support for sampling kernels
#2345 opened Jan 13, 2026 by yzh119 Loading…
29
feat: add LSE return support to TRT LLM attention kernels
#2332 opened Jan 11, 2026 by yzh119 Loading…
1
5
feat: expose swizzled_input_sf parameter for CUTLASS fused MOE
#2330 opened Jan 11, 2026 by yzh119 Loading…
1
6
[wip] feat: add bias support to TGV and CUTLASS BF16 GEMM
#2329 opened Jan 11, 2026 by yzh119 Loading…
1
6
feat: add batch_invariant option to trtllm decode functions
#2321 opened Jan 9, 2026 by yzh119 Loading…
1
23
feat: Support Fused MoE non gated Relu2 NVFP4 & FP8 and support Nemotron run-ci
#2304 opened Jan 7, 2026 by amitz-nv Loading…
3 of 5 tasks
51
[Perf][Feature] Add SM103-specific schedulers for NVFP4 CUTLASS kernels v0.6.2
#2303 opened Jan 7, 2026 by LopezCastroRoberto Loading…
22
chore: Update XFails Report automated maintenance testing
#2287 opened Jan 5, 2026 by flashinfer-bot Loading…
4
ProTip! Filter pull requests by the default branch with base:main.