pytorch

436edc5a - [ONNX] Retire 'DynamoOptimizeExporter' (#99202)

[ONNX] Retire 'DynamoOptimizeExporter' (#99202) <!-- copilot:all --> ### <samp>🤖 Generated by Copilot at f2ccd03</samp> ### Summary 🗑️📝🛠️ <!-- 1. 🗑️ - This emoji represents the removal of unused or unnecessary code, such as the class `DynamoOptimizeExporter` and some imports and decorators. 2. 📝 - This emoji represents the improvement of code readability and consistency, such as replacing `skip_fx_exporters` with `xfail` and using more descriptive names for the FX exporters. 3. 🛠️ - This emoji represents the simplification and refactoring of the code, such as removing some FX exporters and reducing the number of arguments and conditions in the tests. --> Removed unused code and simplified test logic for FX to ONNX conversion. This involved removing `skip_fx_exporters` and `DynamoOptimizeExporter`, and using `xfail` instead of `skip_fx_exporters` in `pytorch_test_common.py` and `test_fx_to_onnx_with_onnxruntime.py`. > _Some FX exporters were not in use_ > _So they were removed without excuse_ > _The tests were updated_ > _With `xfail` annotated_ > _To make the ONNX logic more smooth_ ### Walkthrough * Remove unused imports of `Mapping`, `Type`, and `exporter` from `test/onnx/pytorch_test_common.py` ([link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-26ce853445bf331686abb33393ee166726923ce36aa2a8de98ac7a2e3bc5a6d8L9-R9), [link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-26ce853445bf331686abb33393ee166726923ce36aa2a8de98ac7a2e3bc5a6d8L16-R16)) * Replace custom `skip_fx_exporters` function with standard `xfail` decorator in `test/onnx/pytorch_test_common.py` and `test/onnx/test_fx_to_onnx_with_onnxruntime.py` to simplify test skipping logic and mark tests as expected to fail ([link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-26ce853445bf331686abb33393ee166726923ce36aa2a8de98ac7a2e3bc5a6d8L209-R220), [link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-c8fa56eefd7f98fb4f9739d57df57f02ede77e28528133736010a6d06651ebcbL319-R288), [link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-c8fa56eefd7f98fb4f9739d57df57f02ede77e28528133736010a6d06651ebcbL375-R343), [link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-c8fa56eefd7f98fb4f9739d57df57f02ede77e28528133736010a6d06651ebcbL619-R563), [link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-c8fa56eefd7f98fb4f9739d57df57f02ede77e28528133736010a6d06651ebcbL721-R656), [link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-c8fa56eefd7f98fb4f9739d57df57f02ede77e28528133736010a6d06651ebcbL788-R718)) * Remove unused `DynamoOptimizeExporter` class from `torch/onnx/_internal/fx/dynamo_exporter.py` and remove references to it in `test/onnx/test_fx_to_onnx_with_onnxruntime.py` to simplify FX exporter logic and remove unused code ([link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-3ecf10bc5f6eb95a19441118cb947bd44766dc5eb9b26346f922759bb0f8c9f2L16-L85), [link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-c8fa56eefd7f98fb4f9739d57df57f02ede77e28528133736010a6d06651ebcbL37-R37), [link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-c8fa56eefd7f98fb4f9739d57df57f02ede77e28528133736010a6d06651ebcbL411-L415), [link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-c8fa56eefd7f98fb4f9739d57df57f02ede77e28528133736010a6d06651ebcbL452-L454)) * Remove unused variables and parameters related to different FX exporters in `test/onnx/test_fx_to_onnx_with_onnxruntime.py` and use `torch.onnx.dynamo_export` directly to simplify code ([link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-c8fa56eefd7f98fb4f9739d57df57f02ede77e28528133736010a6d06651ebcbL50-R47), [link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-c8fa56eefd7f98fb4f9739d57df57f02ede77e28528133736010a6d06651ebcbL191-R188), [link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-c8fa56eefd7f98fb4f9739d57df57f02ede77e28528133736010a6d06651ebcbL245-R224), [link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-c8fa56eefd7f98fb4f9739d57df57f02ede77e28528133736010a6d06651ebcbL265-R237), [link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-c8fa56eefd7f98fb4f9739d57df57f02ede77e28528133736010a6d06651ebcbL279), [link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-c8fa56eefd7f98fb4f9739d57df57f02ede77e28528133736010a6d06651ebcbL296)) * Replace `skip` decorators with `xfail` decorators in `test/onnx/test_fx_to_onnx_with_onnxruntime.py` to mark tests as expected to fail instead of skipping them unconditionally ([link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-c8fa56eefd7f98fb4f9739d57df57f02ede77e28528133736010a6d06651ebcbL524-R471), [link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-c8fa56eefd7f98fb4f9739d57df57f02ede77e28528133736010a6d06651ebcbL665-R600), [link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-c8fa56eefd7f98fb4f9739d57df57f02ede77e28528133736010a6d06651ebcbL748-R675), [link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-c8fa56eefd7f98fb4f9739d57df57f02ede77e28528133736010a6d06651ebcbL767-R696)) * Replace `skip_fx_exporters` decorator with `skip_dynamic_fx_test` decorator in `test/onnx/test_fx_to_onnx_with_onnxruntime.py` to skip tests only for dynamic shapes instead of a specific FX exporter ([link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-c8fa56eefd7f98fb4f9739d57df57f02ede77e28528133736010a6d06651ebcbL591-R541), [link](https://github.com/pytorch/pytorch/pull/99202/files?diff=unified&w=0#diff-c8fa56eefd7f98fb4f9739d57df57f02ede77e28528133736010a6d06651ebcbL831-R761)) Pull Request resolved: https://github.com/pytorch/pytorch/pull/99202 Approved by: https://github.com/abock