Default Branch

23ba1065a7 · Update README.md (#1522) · Updated 2026-10-11 17:26:02 +00:00

Branches

57d5ff4cdd · Restore benchmarks after gh-deploy · Updated 2026-10-05 23:46:22 +00:00    linkedin

979
116

ebc9afba0d · Merge branch 'main' into vaibhav/cutile-flce-chunking-triton-policy · Updated 2026-09-14 20:48:13 +00:00    linkedin

65
5

18692d4f9c · feat(cutile): chunk FLCE with cuBLAS gradient GEMMs · Updated 2026-09-10 08:17:00 +00:00    linkedin

70
3

c7908aa662 · feat(flce): expose explicit token chunk sizes · Updated 2026-09-10 07:58:46 +00:00    linkedin

70
2

3eddb174e0 · perf(flce): accumulate low-precision Triton dW with addmm · Updated 2026-09-10 06:36:46 +00:00    linkedin

71
2

1aac00f870 · feat(cutedsl): fused SwiGLU reads stacked/separate weights without pre-packing · Updated 2026-09-09 23:03:08 +00:00    linkedin

77
2

b4d7611759 · SwiGLU backend comparison: refresh after cuTile cliff + CuTe DSL graph fixes · Updated 2026-08-26 20:04:10 +00:00    linkedin

99
3

2f46792776 · Add RoPE backend comparison: Triton vs cuTile vs CuTe DSL · Updated 2026-08-21 22:29:09 +00:00    linkedin

106
1

60e7955c1e · Merge branch 'main' into fix/rope-dead-constexpr-batch-size-dynamic-shapes · Updated 2026-08-07 08:48:09 +00:00    linkedin

134
2

3ef8ab5cd6 · Merge branch 'main' into justinhh4/cutedsl-cross-entropy · Updated 2026-07-03 06:12:19 +00:00    linkedin

155
4

7528b17ba7 · [Megatron] Add fused linear + vocab-parallel cross-entropy (VP-FLCE) · Updated 2026-06-17 06:51:00 +00:00    linkedin

166
1

101f40523c · Merge branch 'main' into nemotron-relu-squared-patch · Updated 2026-03-27 22:47:52 +00:00    linkedin

229
3

a75f34497c · vjindal/testing · Updated 2026-02-27 00:47:28 +00:00    linkedin

274
1

a9bd4f7699 · layer norm opt · Updated 2026-02-25 09:08:49 +00:00    linkedin

274
1

e2df2ae3c3 · optimize jsd · Updated 2026-02-25 07:54:02 +00:00    linkedin

274
1

732b0b179a · Merge queue test, should not be merged. · Updated 2026-02-12 17:51:30 +00:00    linkedin

281
1

edb5ed51f5 · fix typo · Updated 2026-02-07 11:19:30 +00:00    linkedin

296
2

018994adfd · Merge branch 'main' into upgrade-github-actions-node24 · Updated 2026-01-11 19:31:17 +00:00    linkedin

333
2

7167104f6a · Merge branch 'main' into tcc/bench_vis · Updated 2025-12-11 22:08:03 +00:00    linkedin

352
6

8efda76b19 · Update import path for ops in liger_kernel/transformers/* · Updated 2025-12-11 17:06:32 +00:00    linkedin

352
2