Emre Albayrak 2e56847d72 Add Snake activation functor for EVT (#3184) ...

Introduces cutlass::epilogue::thread::Snake, a two-operand activation
functor implementing Snake_a(x) = x + (1/a) * sin^2(a*x) from
Ziyin et al. 2020 (arXiv:2006.08195). The per-channel learnable
frequency `a` flows through an EVT child (e.g. Sm90RowBroadcast),
composing into Sm90EVT<Sm90Compute<Snake, ...>, x_node, alpha_node>
for fused GEMM+Snake epilogues used in neural vocoders.

Adds unit tests in test/unit/epilogue/thread/activation.cu covering
f32 and bf16 paths, validated against float64 reference goldens.

Closes #3141

2026-05-11 22:09:53 -04:00

..

examples/CuTeDSL

[CuTeDSL] Fix loop carried target scope (#3200)

2026-05-11 16:02:26 +08:00

python

v4.3.5 update. (#2934)

2026-01-08 15:02:56 -05:00

self_contained_includes

v4.4 update. (#2979)

2026-01-24 11:46:17 -05:00

unit

Add Snake activation functor for EVT (#3184)

2026-05-11 22:09:53 -04:00

utils

[Hopper CuTeDSL] Add grouped GEMM persistent kernel and tests (#3091)

2026-03-18 00:40:15 -04:00

CMakeLists.txt

v4.3.5 update. (#2934)

2026-01-08 15:02:56 -05:00