onnxruntime/orttraining/orttraining/python/training/ortmodule
pengwa b9d80131a7
Save optimized pre_grad graph once ready (#16816)
### Save optimized pre_grad graph once it's ready

`graph_builder.build()` did two things for training: 1. optimized
forward graph, e.g. pre_grad graph optimization. 2. build gradient
graph.

Originally after `graph_builder.build()` completed, pre_graph graph is
saved. While if pre_grad graph optimization completed, but fail during
gradient graph build, we still cannot get pre_grad graph to investigate.

This PR made the change once pre_grad graph is ready, we save it (if
save_model is enabled) in C++ backend.
2023-08-02 14:05:26 +08:00
..
experimental Disable PERF* rules in ruff to allow better readability (#16834) 2023-07-25 15:38:22 -07:00
torch_cpp_extensions Disable PERF* rules in ruff to allow better readability (#16834) 2023-07-25 15:38:22 -07:00
__init__.py Triton Codegen for ORTModule (#15831) 2023-07-13 18:17:58 +08:00
_custom_autograd_function.py Manage ORTModule configurations consistently (#16396) 2023-06-27 19:19:36 +08:00
_custom_autograd_function_exporter.py Consolidate ORTModule logging (#16078) 2023-06-01 10:09:12 +08:00
_custom_autograd_function_runner.py Adopt linrtunner as the linting tool - take 2 (#15085) 2023-03-24 15:29:03 -07:00
_custom_gradient_registry.py [Better Engineering] Bump ruff to 0.0.278 and fix new lint errors (#16789) 2023-07-21 12:53:41 -07:00
_custom_op_symbolic_registry.py [Better Engineering] Bump ruff to 0.0.278 and fix new lint errors (#16789) 2023-07-21 12:53:41 -07:00
_execution_agent.py Adopt linrtunner as the linting tool - take 2 (#15085) 2023-03-24 15:29:03 -07:00
_fallback.py Consolidate ORTModule logging (#16078) 2023-06-01 10:09:12 +08:00
_fallback_exceptions.py Adopt linrtunner as the linting tool - take 2 (#15085) 2023-03-24 15:29:03 -07:00
_gradient_accumulation_manager.py Adopt linrtunner as the linting tool - take 2 (#15085) 2023-03-24 15:29:03 -07:00
_graph_execution_interface.py Adopt linrtunner as the linting tool - take 2 (#15085) 2023-03-24 15:29:03 -07:00
_graph_execution_manager.py Save optimized pre_grad graph once ready (#16816) 2023-08-02 14:05:26 +08:00
_graph_execution_manager_factory.py ORTModule log clean up (#16795) 2023-07-26 12:42:50 +08:00
_inference_manager.py ORTModule log clean up (#16795) 2023-07-26 12:42:50 +08:00
_io.py [Better Engineering] Bump ruff to 0.0.278 and fix new lint errors (#16789) 2023-07-21 12:53:41 -07:00
_logger.py ORTModule log clean up (#16795) 2023-07-26 12:42:50 +08:00
_onnx_models.py Save optimized pre_grad graph once ready (#16816) 2023-08-02 14:05:26 +08:00
_runtime_inspector.py ORTModule log clean up (#16795) 2023-07-26 12:42:50 +08:00
_torch_module_factory.py Manage ORTModule configurations consistently (#16396) 2023-06-27 19:19:36 +08:00
_torch_module_interface.py Adopt linrtunner as the linting tool - take 2 (#15085) 2023-03-24 15:29:03 -07:00
_torch_module_ort.py ORTModule log clean up (#16795) 2023-07-26 12:42:50 +08:00
_torch_module_pytorch.py Adopt linrtunner as the linting tool - take 2 (#15085) 2023-03-24 15:29:03 -07:00
_training_manager.py Save optimized pre_grad graph once ready (#16816) 2023-08-02 14:05:26 +08:00
_utils.py ORTModule log clean up (#16795) 2023-07-26 12:42:50 +08:00
graph_transformer_registry.py [Better Engineering] Bump ruff to 0.0.278 and fix new lint errors (#16789) 2023-07-21 12:53:41 -07:00
options.py ORT Cache (#16744) 2023-07-27 09:00:43 -07:00
ortmodule.py ORTModule log clean up (#16795) 2023-07-26 12:42:50 +08:00