- Extend support:
- quantized::conv1d
- quantized::conv3d
- quantized::conv3d_relu
- quantized::conv_transpose1d
- quantized::conv_transpose2d
- quantized::conv_transpose3d
- Note: quantized::{conv1d_relu,conv2d,conv2d_relu} already supported.
- To support this, quantization unpacking added for:
- conv1d
- conv_transpose1d
- conv_transpose2d
- conv_transpose3d
- Note: conv3d/conv3d_relu already had weights unpacking set up, even though it didn't have torch.onnx support.
- Add tests.
- The 3D tests will fail if run with the qnnpack backend (e.g., on Apple silicon Mac), so added decorator skipIfQuantizationBackendQNNPack.
- Minor fix in `aten/src/ATen/native/quantized/cpu/qconv.cpp` for 3D convolutions (triggered by added tests).
Fixes#102747
Pull Request resolved: https://github.com/pytorch/pytorch/pull/102759
Approved by: https://github.com/BowenBao, https://github.com/thiagocrepaldi, https://github.com/kit1980
Summary
* Introduce input/output adapter. Due to design differences, input/output format
between PyTorch model and exported ONNX model are often not the same. E.g., `None`
inputs are allowed for PyTorch model, but are not supported by ONNX. Nested constructs
of tensors are allowed for PyTorch model, but only flattened tensors are supported by ONNX,
etc. The new input/output adapter is exported with the model. Providing an interface to
automatically convert and validate inputs/outputs format.
* As suggested by #98251,
provide extension for unwrapping user defined python classes for `dynamo.export` based
exporter. Unblock huggingface models.
* Re-wire tests to run through `DynamoExporter` w/ `dynamo_export` api. Kept
`DynamoOptimizeExporter` in the tests for now for coverage of this change.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/98421
Approved by: https://github.com/justinchuby, https://github.com/titaiwangms, https://github.com/thiagocrepaldi
Fixes#97728Fixes#98622
Fixes https://github.com/microsoft/onnx-script/issues/393
Provide op_level_debug in exporter which creates randomnied torch.Tensor based on FakeTensorProp real shape as inputs of both torch ops and ONNX symbolic function. The PR leverages on Transformer class to create a new fx.Graph, but shares the same Module with the original one to save memory.
The test is different from [op_correctness_test.py](https://github.com/microsoft/onnx-script/blob/main/onnxscript/tests/function_libs/torch_aten/ops_correctness_test.py) as op_level_debug generating real tensors based on the fake tensors in the model.
Limitation:
1. Some of the trace_only function is not supported due to lack of param_schema which leads to arg/kwargs wronly split and ndarray wrapping. (WARNINGS in SARIF)
2. The ops with dim/indices (INT64) is not supported that they need the information(shape) from other input args. (WARNINGS in SARIF)
3. sym_size and built-in ops are not supported.
4. op_level_debug only labels results in SARIF. It doesn't stop exporter.
5. Introduce ONNX owning FakeTensorProp supports int/float/bool
6. parametrized op_level_debug and dynamic_shapes into FX tests
Pull Request resolved: https://github.com/pytorch/pytorch/pull/97494
Approved by: https://github.com/justinchuby, https://github.com/BowenBao
- Rename `test_pytorch_common` -> `pytorch_test_common`, `test_onnx_common` -> `onnx_test_common`, removing the test_ prefix to show that the files are not test cases
- Remove import * in `test_pytorch_common` and adjust to import from `testing._internal.common_utils` (where functions are actually defined) instead
- Import modules only in `test_pytorch_onnx_onnxruntime` (too many to handle in a single PR in other tests) (The skips are exceptions)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/81141
Approved by: https://github.com/BowenBao
