Default Branch

1076f97059 · delete autocast, an unnecessary thorn in my side, manage dtypes directly · Updated 2026-03-04 23:55:30 +00:00

Branches

moe

5422d3a132 · make sure to use active params in scaling laws · Updated 2026-02-19 02:46:36 +00:00

9
4

50bea28ef9 · also add readme mention of the cpu mps changes · Updated 2025-10-21 17:24:48 +00:00

297
0
Included

69b1ed245e · also add base_train change example for how to swap LinearFP8 · Updated 2026-01-13 17:08:10 +00:00

125
2