Wei-Sheng Chin c8de3eaac6 Update DORT to follow PyTorch changes (#16394) ... Fix #16355. The root cause change in PyTorch is [#103302](https://github.com/pytorch/pytorch/pull/103302), which seem blocking calling make_fx inside a dynamo backend. Changes: 1. Move decomposition to `register_backend.py`, so we don't have to call `make_fx` inside DORT, which triggers a bunch of new exceptions. 2. Remove shape inference based on FakeTensorProp since the FX graph received from dynamo contains all shapes now. 3. Fix a macro bug so that DORT can build without CUDA. Before (3), ``` #if defined(USE_CUDA) || defined(USE_ROCM) virtual PhiloxGenerator& PhiloxGenerator__Default() = 0; #ifdef ENABLE_TRAINING_TORCH_INTEROP ... #endif #endif ``` After (3), ``` #if defined(USE_CUDA) || defined(USE_ROCM) virtual PhiloxGenerator& PhiloxGenerator__Default() = 0; #endif #ifdef ENABLE_TRAINING_TORCH_INTEROP ... #endif ``` The later one looks better since the `ENABLE_TRAINING_TORCH_INTEROP` is for Python bridge code, not for random-number-generating kernels `PhiloxGenerator`.		2023-06-20 12:06:50 -07:00
..
orttraining	Update DORT to follow PyTorch changes (#16394)	2023-06-20 12:06:50 -07:00
pytorch_frontend_examples	Enable pylint and numpy rules (#15218)	2023-03-27 20:37:53 -07:00
tools	[ROCm] reduce batch size to fix CI error (#15714)	2023-05-16 13:10:02 +08:00