Skip to content

Pull requests: InfiniTensor/InfiniOps

New pull request New
80 Open 891 Closed
80 Open 891 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: add LightningAttentionInfinilm for MiniMax linear attention
#995 opened Sep 19, 2026 by y258dd Loading…
5 of 19 tasks
feat(metax): add moe_topk_softmax operator for MoE top-k gating
#993 opened Sep 18, 2026 by rubik-hua Loading…
feat(metax): add moe_fused_dense operator for fused MoE expert forward
#992 opened Sep 18, 2026 by rubik-hua Loading…
feat(metax): add moe_align operator for MoE token alignment
#991 opened Sep 18, 2026 by rubik-hua Loading…
fix(iluvatar): correct HD128 paged prefill dispatch
#990 opened Sep 18, 2026 by gongchensu Contributor Loading…
2 of 19 tasks
@gongchensu
feat(moore): call MATE flash attention natively
#987 opened Sep 15, 2026 by voltjia Collaborator Loading…
4 tasks
fix(ascend): detect exact SoC version
#986 opened Sep 14, 2026 by baominghelly Contributor Loading…
3 of 18 tasks
feat(moore): support flash-attn via MooreThreads/mate v0.2.5
#983 opened Sep 9, 2026 by spike-zhu Contributor Loading…
2 of 19 tasks
@spike-zhu
feat(cambricon): add convolution provider
#967 opened Sep 2, 2026 by baominghelly Contributor • Draft
3 of 19 tasks
feat(cambricon): add topk softmax provider
#958 opened Aug 31, 2026 by baominghelly Contributor • Draft
2 of 19 tasks
fix: preserve implementation lookups under optimization
#929 opened Aug 13, 2026 by voltjia Collaborator Loading…
9 of 19 tasks
perf: optimize cache for tensor conversion
#928 opened Aug 13, 2026 by DockeyMint Loading…
3 of 19 tasks
feat(nvidia): add top-k top-p sampling provider
#922 opened Aug 11, 2026 by voltjia Collaborator • Draft
3 of 19 tasks
refactor: simplify operator cache fast path
#920 opened Aug 11, 2026 by voltjia Collaborator • Draft
10 of 19 tasks
feat(nvidia): add Argmax provider
#918 opened Aug 10, 2026 by voltjia Collaborator Loading…
3 of 19 tasks
fix(rms_norm): support optional weight
#916 opened Aug 9, 2026 by voltjia Collaborator • Draft
9 of 19 tasks
feat(nvidia): link vLLM fused_marlin_moe
#914 opened Aug 8, 2026 by voltjia Collaborator • Draft
5 of 19 tasks
fix(linked): support multiple operator schemas
#913 opened Aug 8, 2026 by voltjia Collaborator • Draft
5 of 19 tasks
fix: support optional C in Gemm
#870 opened Aug 4, 2026 by voltjia Collaborator • Draft
11 of 20 tasks
perf: avoid redundant operator cache lookups
#858 opened Jul 30, 2026 by baominghelly Contributor Loading…
10 of 19 tasks
perf(torch): reuse ATen wrappers in generated operators
#816 opened Jul 24, 2026 by voltjia Collaborator • Draft
11 of 19 tasks
perf: reuse ATen tensors and optimize PyTorch fallback wrappers
#798 opened Jul 13, 2026 by baominghelly Contributor • Draft
13 of 19 tasks
feat(ascend): add rotary_embedding operator
#786 opened Jun 30, 2026 by zhangyue207 Contributor • Draft
feat(ascend): add add_rms_norm operator
#785 opened Jun 30, 2026 by zhangyue207 Contributor • Draft
feat(ascend): add rms_norm operator
#784 opened Jun 30, 2026 by zhangyue207 Contributor • Draft
ProTip! Find all pull requests that aren't related to any open issues with -linked:issue.