swift-coreml-diffusers

Author	SHA1	Message	Date
Pedro Cuenca	3a4e04ab1a	Decide attention and ANE on hardware capabilities. For chips with >= 8 performance cores, we select CPU+GPU (original attention). Otherwise we select CPU+ANE (split einsum). Some computers (M1 Pro, 16 core GPU) might yield slightly better performance using CPU+GPU+ANE with SPLIT_EINSUM.	2 years ago
Pedro Cuenca	8796695928	Advanced Settings: ANE (#28 ) * Preparing to allow users to override inference settings. * ANE setting. * Do not show advanced settings if ane is not available.	2 years ago
Pedro Cuenca	e2cc04fce3	Implement ModelInfo.Equatable.	2 years ago
Pedro Cuenca	d3c145f219	Add v1-4 and ofa-sys v0 models.	2 years ago
Pedro Cuenca	06ce6e56b5	macOS UI (#12 ) Native macOS UI Co-authored-by: Cyril Zakka <1841186+cyrilzakka@users.noreply.github.com>	2 years ago
Pedro Cuenca	3867b905b3	Add simple metadata for a couple of models.	3 years ago