Vincent Wang 3654a5d60e Register Custom Symbolic of torch.einsum for ORTModule (#9590) ... * register custom symbolic for einsum * bugfix for case needs permute at the end * refactor * refactor equation parser * support new case, use ReduceProd * optimize perf and graph * remove some Gather node * add more ut, fix gemm trans fusion		2021-11-18 10:13:58 +08:00
..
deprecated	[python api] align api with other language bindings' treatment of explicit provider registrations. enforce use of providers param in python InferenceSession when execution providers other than default CPU are enabled. (#9712)	2021-11-10 12:17:53 -08:00
training	Register Custom Symbolic of torch.einsum for ORTModule (#9590)	2021-11-18 10:13:58 +08:00
checkpointing_utils.py	Improve checkpointing for Zero stage 1 (#5478)	2020-12-07 09:16:01 -08:00
ort_trainer.py	Include pytorch_export_contrib_ops in inference builds (#8878)	2021-09-02 14:26:58 -07:00
orttraining_pybind_common.h	resolve the provider options before create training session in orttrainer (#9199)	2021-10-12 09:30:45 -07:00
orttraining_pybind_state.cc	Update default cast propagation strategy from None to FloodFill (#9713)	2021-11-16 13:15:57 -08:00
orttraining_python_module.cc	[ROCm] static re-hipify of CUDA EP to ROCm EP, now a shared provider (#8877)	2021-10-14 15:15:51 -07:00
pt_patch.py	Liqun/bert pretrain2 (#5327)	2020-10-01 11:01:26 -07:00