Skip to content

Pull requests: linkedin/Liger-Kernel

New pull request New
95 Open 938 Closed
95 Open 938 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

[NPU] Pin torch/torch_npu=2.9.0, triton-ascend=3.2.2 and cann=9.1.0 for NPU
#1392 opened Aug 17, 2026 by zheliuyu Contributor • Draft
3 tasks done
Add FP32 accumulation option to fused scaled cross entropy
#1391 opened Aug 17, 2026 by yueyiming2009 Collaborator Loading…
4 of 7 tasks
Add Liger Kernel support for Muse Glimmer
#1390 opened Aug 16, 2026 by buffett0323 Contributor Loading…
3 tasks done
4
fix: restore cross-entropy state in revert_liger_kernel helpers
#1389 opened Aug 16, 2026 by atirna Loading…
2 of 4 tasks
1
Add fused DeepSeek-V4 partial interleaved RoPE
#1387 opened Aug 16, 2026 by vedavyas2727 Loading…
2 of 3 tasks
Fix broken star history chart
#1385 opened Aug 15, 2026 by Dessalines39394 Loading…
Add fused CE + TV distance kernel
#1384 opened Aug 15, 2026 by kashif Contributor Loading…
1 of 3 tasks
benchmark: add RTX 5060 Ti RMSNorm results
#1382 opened Aug 14, 2026 by alifurkanstahl • Draft
Add DeepSeek V3 support
#1379 opened Aug 13, 2026 by SamJSui Loading…
3 tasks done
1
Fix group norm backward for inputs with more than three dimensions
#1376 opened Aug 11, 2026 by truong-v Loading…
1 of 3 tasks
1
Fix TVD gradient at p == q
#1374 opened Aug 11, 2026 by truong-v Loading…
1 of 3 tasks
1
fix(cutedsl): preserve FP32 CE output precision
#1372 opened Aug 11, 2026 by justinhh4 Contributor Loading…
feat: add Megatron fused linear cross entropy backends
#1371 opened Aug 11, 2026 by justinhh4 Contributor Loading…
[Ascend] Fix missing accum_dtype param in fused_linear_jsd forward
#1367 opened Aug 11, 2026 by TecJesh Loading…
2 of 3 tasks
Add SM100 CuTe DSL GRPO fast path
#1363 opened Aug 10, 2026 by Celaena24 Contributor Loading…
2
Fix global LayerNorm mutation in model patching
#1360 opened Aug 10, 2026 by vedavyas2727 Loading…
1 of 3 tasks
1
feat: add LigerMLP module
#1357 opened Aug 9, 2026 by Pearblossom-M Loading…
2 of 3 tasks
24
Fix ZeroDivisionError in fused linear cross entropy when weight vocab dim is 0
#1353 opened Aug 7, 2026 by nileshpatil6 Loading…
1 of 3 tasks
1
1
fix(rope): don't mark the unused batch-size param as tl.constexpr
#1351 opened Aug 7, 2026 by vaibhavjindal Collaborator • Draft
1 of 3 tasks
feat: add support for LFM models
#1345 opened Aug 5, 2026 by kadirnar Loading…
Fix AttnRes silently dropping blocks past 32
#1344 opened Aug 5, 2026 by adityasingh2400 Loading…
2
Cast tl.program_id(0) to int64 in llama4_rope and qwen2vl_mrope
#1343 opened Aug 5, 2026 by akshansh47 Loading…
1 task
1
Cast tl.program_id to int64 in llama4_rope and qwen2vl_mrope
#1342 opened Aug 5, 2026 by adityasingh2400 Loading…
1 of 3 tasks
1
5
Add SM100 fused linear SwiGLU
#1340 opened Aug 4, 2026 by justinhh4 Contributor Loading…
fix: torch.compile loss corruption at ignore_index rows (+ perf: drop BT x V loss buffer in FLJSD)
#1338 opened Aug 4, 2026 by KyleMylonakisProtopia Contributor Loading…
3 tasks done
1
ProTip! Exclude everything labeled bug with -label:bug.