Files

torchao

Feb 24, 2025

9685b25 · Feb 24, 2025

Name	Name	Last commit message	Last commit date
parent directory ..
_models	_models	promote blocksparse from prototype, make it faster (pytorch#1734 )	Feb 19, 2025
core	core	[bc-breaking] enable direct configuration in quantize_ (pytorch#1595 )	Feb 14, 2025
csrc	csrc	Make TorchAO cpp/Python extension	Feb 18, 2025
dtypes	dtypes	add nf4 metadata	Feb 24, 2025
experimental	experimental	Add ukernel selection logic + clean up KleidiAI integration (pytorch#…	Feb 20, 2025
float8	float8	float8 training: add README.md entry for rowwise scaling (pytorch#1733 )	Feb 19, 2025
kernel	kernel	promote blocksparse from prototype, make it faster (pytorch#1734 )	Feb 19, 2025
prototype	prototype	promote blocksparse from prototype, make it faster (pytorch#1734 )	Feb 19, 2025
quantization	quantization	Make FakeQuantizer expose useful config details (pytorch#1717 )	Feb 19, 2025
sparsity	sparsity	promote blocksparse from prototype, make it faster (pytorch#1734 )	Feb 19, 2025
testing	testing	for now, delete the float8-all-gather-only functionality from float8 … (	Dec 20, 2024
__init__.py	__init__.py	Make TorchAO cpp/Python extension	Feb 18, 2025
_executorch_ops.py	_executorch_ops.py	Lint fixes torchao files (pytorch#1366 )	Dec 2, 2024
ops.py	ops.py	promote blocksparse from prototype, make it faster (pytorch#1734 )	Feb 19, 2025
utils.py	utils.py	Add boiler plate code to Tensor subclass (pytorch#1663 )	Feb 7, 2025