onnxruntime/orttraining/orttraining/python/training
pengwa d03e477b90
Fix missing subgraph candidates for recompute (#19077)
### Fix missing subgraph candidates for recompute

For subgraphs for example `MatMul+Transpose+Reshape`, since the ending
node is a Reshape, in ORT, it is reusing input buffers.

Currently, the subgraph detection logic has defect, as a result, those
subgraphs will be missing as recompute candidates.

Also append a few more node types for recompute support. 

TODO: add unit test later. This PR is needed for a customer model now.
2024-01-11 12:50:55 +08:00
..
amp [Better Engineering] Bump ruff to 0.0.278 and fix new lint errors (#16789) 2023-07-21 12:53:41 -07:00
api [On-Device Training] Expose Parameters through the Training API (#17364) 2023-09-25 20:03:24 -07:00
experimental Manage ORTModule configurations consistently (#16396) 2023-06-27 19:19:36 +08:00
onnxblock [Linter] Bump ruff and remove pylint (#17797) 2023-10-05 21:07:33 -07:00
optim FP16 optimizer automatically detect DeepSpeed compatibility (#18084) 2023-10-25 15:11:02 +08:00
ort_triton [ORTModule] Remove Unused Arguments from Generated Triton Code (#18636) 2023-11-30 18:32:36 +08:00
ortmodule Fix missing subgraph candidates for recompute (#19077) 2024-01-11 12:50:55 +08:00
utils Minor fixes (#18949) 2023-12-28 20:01:06 +08:00
__init__.py Removed all the deprecated python training code and related tests and utils (#18333) 2023-11-17 18:19:21 -08:00
_utils.py Removed all the deprecated python training code and related tests and utils (#18333) 2023-11-17 18:19:21 -08:00
artifacts.py Fix opset version of the optimizer in function generate_artifacts (#18300) 2023-11-22 09:15:11 -08:00