Skip to content

Actions: linkedin/Liger-Kernel

Actions

Publish Liger Kernel Nightly

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
publish-nightly.yml
367 workflow runs
367 workflow runs

Filter by Event

Filter by Status

Filter by Branch

Filter by Actor

Update checkstyle and fix the format issue (#1071) Publish Liger Kernel Nightly #670: Commit 1fdb668 pushed by Tcc0403
37s main
main
37s
Add CISPO loss type support for LigerFusedLinearGRPOLoss (#1054) Publish Liger Kernel Nightly #669: Commit 83cdcf8 pushed by Tcc0403
35s main
main
35s
New version release v0.6.5 (#1063) Publish Liger Kernel Nightly #668: Commit 81f932a pushed by vaibhavjindal
30s main
main
30s
Unify NPU vector core count helpers (#1052) Publish Liger Kernel Nightly #667: Commit bb88671 pushed by Tcc0403
35s main
main
35s
Set transformers upper bound (#1046) Publish Liger Kernel Nightly #666: Commit a5be02e pushed by vaibhavjindal
35s main
main
35s
Fix memory requirements for benchmark_jsd.py and `benchmark_distill… Publish Liger Kernel Nightly #665: Commit 7656bcb pushed by Tcc0403
32s main
main
32s
[NPU] Add Llama4_rope support on NPU (#1035) Publish Liger Kernel Nightly #664: Commit 0ea0b8f pushed by Tcc0403
36s main
main
36s
[NPU]: optimize rope and mrope implementation (#1041) Publish Liger Kernel Nightly #663: Commit 1f51687 pushed by Tcc0403
36s main
main
36s
Avoid OOM error for benchmark_tvd.py on GPUs with less than 66GB of… Publish Liger Kernel Nightly #662: Commit 7aa3a4b pushed by Tcc0403
32s main
main
32s
[NPU] FIX CE ub overflow on NPU (#1040) Publish Liger Kernel Nightly #661: Commit 9eb9a1e pushed by Tcc0403
36s main
main
36s
Fix benchmark_qwen2vl_mrope.py by enabling new transformers API for… Publish Liger Kernel Nightly #660: Commit 4444fdf pushed by Tcc0403
38s main
main
38s
Workaround for OOM error on benchmark_jsd (#1037) Publish Liger Kernel Nightly #659: Commit ab07bfd pushed by Tcc0403
33s main
main
33s
[NPU]: optimize tvd implementation (#1039) Publish Liger Kernel Nightly #658: Commit 7b51e56 pushed by Tcc0403
32s main
main
32s
[NPU]: optimize GEGLU implementation with flatten 1D approach (#1031) Publish Liger Kernel Nightly #657: Commit 224a660 pushed by Tcc0403
32s main
main
32s
[NPU]: use get_soc_spec for UB capacity detection (#1038) Publish Liger Kernel Nightly #656: Commit 15f9338 pushed by Tcc0403
32s main
main
32s
[NPU]: Add NPU support for the embedding (#1028) Publish Liger Kernel Nightly #655: Commit 57e98d3 pushed by Tcc0403
31s main
main
31s
doc: reformat contributing.md for better visualization on github (#1033) Publish Liger Kernel Nightly #654: Commit 393efae pushed by vaibhavjindal
32s main
main
32s
[NPU]: update the native KLDivLoss implementation for comparison. (eg… Publish Liger Kernel Nightly #653: Commit b708f79 pushed by Tcc0403
31s main
main
31s
Add pre-commit config (#1009) Publish Liger Kernel Nightly #652: Commit 559e9a1 pushed by lancerts
33s main
main
33s
Enable benchmark_tvd.py for xpu devices (#1024) Publish Liger Kernel Nightly #651: Commit 25189b8 pushed by Tcc0403
33s main
main
33s
[Refactor]: optimize poly_norm backward kernel pointer handling (#1018) Publish Liger Kernel Nightly #650: Commit 48fb31f pushed by Tcc0403
35s main
main
35s
gemma3 consider loss_kwargs (#1007) Publish Liger Kernel Nightly #649: Commit 3b79375 pushed by Tcc0403
33s main
main
33s
XPU: Enable new grf_mode settings (#1016) Publish Liger Kernel Nightly #648: Commit 71ed8ac pushed by Tcc0403
34s main
main
34s
avoid pointer mutation in layer_norm kernel (#1006) Publish Liger Kernel Nightly #647: Commit 2ce9bdd pushed by Tcc0403
35s main
main
35s
pass param down to LigerFusedLinearCrossEntropyLoss (#1010) Publish Liger Kernel Nightly #646: Commit 41d4bcf pushed by vaibhavjindal
33s main
main
33s