Skip to content

Actions: flashinfer-ai/flashinfer

Actions

Claude Code Review

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
claude-code-review.yml
2,500+ workflow runs
2,500+ workflow runs

Filter by Event

⌥ + click/return to exclude

Filter by Status

⌥ + click/return to exclude

Filter by Branch

⌥ + click/return to exclude

Filter by Actor

⌥ + click/return to exclude
[Fmha] Add head_dim=512 support for trtllm attention kernels Claude Code Review #2757: Pull request #2959 synchronize by yzh119
8s djmmoss:dmoss/trtllm-fmha-head-dim-512
djmmoss:dmoss/trtllm-fmha-head-dim-512
8s
feat: Enable FP8 (E4M3/E5M2) in concat_mla_k for optimize long-context prefill performance and refactor type dispatch for BF16/FP16 Claude Code Review #2756: Pull request #3129 synchronize by qiching
9s qiching:fix/concat-mla-k-fp8-support
qiching:fix/concat-mla-k-fp8-support
9s
feat: add delimiterless multi-item scoring (MaskMode::kMultiItemScoringV2) Claude Code Review #2755: Pull request #2921 synchronize by chanh
Action required chanh:cnguyen/delimiterless-mis-v2
chanh:cnguyen/delimiterless-mis-v2
Action required
[feat] Trtllm-gen Per-token Nvfp4 MoE Claude Code Review #2754: Pull request #3027 synchronize by IwakuraRein
8s IwakuraRein:per-token-fp4
IwakuraRein:per-token-fp4
8s
Vendor CCCL v3.3.2 from GitHub instead of relying on CTK-bundled copy Claude Code Review #2753: Pull request #3091 synchronize by kahyunnam
10s kahyunnam:knam/cccl-submodule
kahyunnam:knam/cccl-submodule
10s
feat(dllm): add Block Extend Attention for Diffusion LLM Claude Code Review #2752: Pull request #2722 synchronize by fdz-1999
Action required fdz-1999:feature/block-extend
fdz-1999:feature/block-extend
Action required
feat(dllm): add Block Extend Attention for Diffusion LLM Claude Code Review #2751: Pull request #2722 synchronize by fdz-1999
Action required fdz-1999:feature/block-extend
fdz-1999:feature/block-extend
Action required
feat(dllm): add Block Extend Attention for Diffusion LLM Claude Code Review #2750: Pull request #2722 synchronize by fdz-1999
Action required fdz-1999:feature/block-extend
fdz-1999:feature/block-extend
Action required
perf(autotuner): replace power-of-2 token buckets with hybrid spacing & fix missing routing_replay_out arg Claude Code Review #2749: Pull request #3115 synchronize by samuellees
12s StudyingShao:jiangs/autotuner-hybrid-bucket-spacing-main
StudyingShao:jiangs/autotuner-hybrid-bucket-spacing-main
12s
feat(dllm): add Block Extend Attention for Diffusion LLM Claude Code Review #2748: Pull request #2722 synchronize by fdz-1999
Action required fdz-1999:feature/block-extend
fdz-1999:feature/block-extend
Action required
perf: optimize MXFP4xBF16 & INT4xFP8 CUTLASS MoE backend for SM90 Claude Code Review #2747: Pull request #3084 synchronize by samuellees
7s samuellees:feat/w4a16-moe-kernel
samuellees:feat/w4a16-moe-kernel
7s
perf: optimize MXFP4xBF16 & INT4xFP8 CUTLASS MoE backend for SM90 Claude Code Review #2746: Pull request #3084 synchronize by samuellees
8s samuellees:feat/w4a16-moe-kernel
samuellees:feat/w4a16-moe-kernel
8s
perf: optimize MXFP4xBF16 & INT4xFP8 CUTLASS MoE backend for SM90 Claude Code Review #2745: Pull request #3084 synchronize by samuellees
8s samuellees:feat/w4a16-moe-kernel
samuellees:feat/w4a16-moe-kernel
8s
perf: optimize MXFP4xBF16 & INT4xFP8 CUTLASS MoE backend for SM90 Claude Code Review #2744: Pull request #3084 synchronize by samuellees
7s samuellees:feat/w4a16-moe-kernel
samuellees:feat/w4a16-moe-kernel
7s
feat: Integrate CuTe DSL FMHA prefill kernels by loading cubin Claude Code Review #2743: Pull request #3039 synchronize by limin2021
12s limin2021:integrate_dsl_cubin_fmha
limin2021:integrate_dsl_cubin_fmha
12s
fix(mla): widen page index to int64_t to avoid 32-bit overflow Claude Code Review #2742: Pull request #3136 opened by Tracin
Action required Tracin:fix_mla
Tracin:fix_mla
Action required
Loosened trtllm_ragged_attention_deepseek shape assertion Claude Code Review #2741: Pull request #3064 synchronize by nvjullin
9s nvjullin:allow-glm-5-shape
nvjullin:allow-glm-5-shape
9s
[fmha-v2] Support HND and NHD paged KV cache layouts with conditional stride handling Claude Code Review #2740: Pull request #2799 synchronize by qsang-nv
8s zhou-yuxin:HND
zhou-yuxin:HND
8s
fix(jit): decouple FLASHINFER_JIT_VERBOSE from debug compilation flags Claude Code Review #2739: Pull request #3135 opened by leonardozcm
Action required leonardozcm:fix/decouple-jit-verbose-from-debug
leonardozcm:fix/decouple-jit-verbose-from-debug
Action required
feat: implement configurable tie_break for filtered topk Claude Code Review #2738: Pull request #3095 synchronize by ziang-and
10s zianglih:topk-tie
zianglih:topk-tie
10s
Implement Gated Delta Rule for sm_120a (Blackwell RTX) Claude Code Review #2737: Pull request #3088 synchronize by guangyunh-nv
8s guangyunh/sm_120a
guangyunh/sm_120a
8s
feat: Add row_starts and dsa_graph_safe to topk Claude Code Review #2736: Pull request #3133 synchronize by ziang-and
Action required zianglih:dsa-graph-safe
zianglih:dsa-graph-safe
Action required
feat: Integrate CuTe DSL FMHA prefill kernels by loading cubin Claude Code Review #2735: Pull request #3039 synchronize by yzh119
7s limin2021:integrate_dsl_cubin_fmha
limin2021:integrate_dsl_cubin_fmha
7s
feat: implement configurable tie_break for filtered topk Claude Code Review #2734: Pull request #3095 synchronize by ziang-and
Action required zianglih:topk-tie
zianglih:topk-tie
Action required
feat: implement configurable tie_break for filtered topk Claude Code Review #2733: Pull request #3095 synchronize by ziang-and
Action required zianglih:topk-tie
zianglih:topk-tie
Action required