COMMITS
/ torchao/float8/fsdp_utils.py August 13, 2025
A
Track API usage (#2706)
andrewor14 committed
July 18, 2025
D
allowlist WeightWithDynamicFloat8CastTensor for deserialization for checkpointing (#2573)
Daniel Vega-Myhre committed
July 12, 2025
J
Rename torchao.float8.Float8Tensor to Float8TrainingTensor (#2535)
Jerry Zhang committed
February 22, 2025
V
delete delayed scaling from torchao.float8 (#1753)
Vasiliy Kuznetsov committed
December 19, 2024
V
refactor: move delayed and static scaling to StatefulFloat8Linear (#1439)
Vasiliy Kuznetsov committed
December 17, 2024
D
Fix error message in float8 FSDP utils (#1423)
Daniel Vega-Myhre committed
December 4, 2024
L
[float8] Allow specifying arbitrary dtype for each tensor (#1378)
Luca Wehrstedt committed
November 11, 2024
A
Ruff Lint fixes (torchao/float8) (#1239)
Apurva Jain committed
October 1, 2024
W
[float8] improve eager numerics for dynamic scales and gets on par with torch.compile (#904)
Wei (Will) Feng committed
September 26, 2024
W
[float8] all-reduce amax on dp mesh instead of global pg (#933)
Wei (Will) Feng committed
September 11, 2024
W
[FSDP2] cast scale to float32 in precompute (#835)
Wei (Will) Feng committed
August 28, 2024
V
float8 training: add static scaling (#760)
Vasiliy Kuznetsov committed
August 22, 2024
M
Specify output dtype to `torch.float32` in `_foreach_norm` (#727)
Masaki Kozuki committed
M
cast `local_scale_tensor` to fp32 for precompute of fp8 dynamic scaling (#713)
Masaki Kozuki committed
August 3, 2024
W
[float8] fix all-gather in 2D with DTensor(WeightWithDynamicFloat8CastTensor) (#590)
Wei (Will) Feng committed
July 30, 2024
V
move float8_experimental to torchao/float8
Vasiliy Kuznetsov committed