George White 6c187246d6 Add support for float8_e4m3fnuz and _e5m2fnuz (#107586) ... This PR relates to the feature in [this feature submission](https://docs.google.com/document/d/1pF2T1xz54IPg1jG7FhykbrpbcJZVelQw0v8vBaoLkfs/edit). It has been based on #104242 which adds similar float8 types. These new types added in this PR are described in the paper at https://arxiv.org/abs/2206.02915. A brief description and comparison of the types with other float8 types can be also found in the [OpenXLA RFC](https://github.com/openxla/stablehlo/blob/main/rfcs/20230321-fp8_fnuz.md). Pull Request resolved: https://github.com/pytorch/pytorch/pull/107586 Approved by: https://github.com/seemethere, https://github.com/malfet		2023-11-15 15:01:11 +00:00
..
ao_migration
bc	[BE] Enable ruff's UP rules and autoformat test/ (#105434)	2023-07-19 20:36:06 +00:00
core	Add support for float8_e4m3fnuz and _e5m2fnuz (#107586)	2023-11-15 15:01:11 +00:00
eager	[pytorch][ao] Add torch.matmul in FloatFunctional/QFunctional (#106831)	2023-08-10 22:43:36 +00:00
fx	Back out "Enable pickling model prepared with QAT qconfig" (#110392)	2023-10-05 14:41:00 +00:00
jit	Revert "Remove deprecated fbgemm operators (#104535)"	2023-10-25 16:34:16 +00:00
pt2e	[quant][pt2][be] Refactor QAT tests for future patterns (#113658)	2023-11-15 02:17:13 +00:00
serialized
__init__.py