Summary:
Many constructors like `torch.zeros` or `torch.randn` didn't support
size tracing correctly which is fixed by this pass. Same issue has been
fixed in legacy tensor constructors.
Additionally, new tensor constructors, which do not participate in
tracing (most notably `torch.tensor`, `torch.as_tensor` and
`torch.from_numpy`) raise a warning when they are used.
Finally, entering a traceable operation disables the tracing in its body.
This is needed because
zdevito
Pull Request resolved: https://github.com/pytorch/pytorch/pull/11288
Reviewed By: ezyang
Differential Revision: D9751183
Pulled By: apaszke
fbshipit-source-id: 51444a39d76a3e164adc396c432fd5ee3c8d5f7f
Summary:
Operators like aten::chunk used to return a number of tensors, but
now return a list. To make it easier to do shape prop through
aten::chunk and fuse it, I've also introduced prim::ConstantChunk,
which behaves like the previous implementation (has a variable length
output list).
The downside of this PR is that the introduction of more lists to the IR causes the LSTM and MiLSTM graphs to be considered as non-differentiable by the graph executor. I verified that they are still optimize correctly, and my next patch (that changes how the specializations/differentiation works) will restore those.
zdevito
Pull Request resolved: https://github.com/pytorch/pytorch/pull/10949
Reviewed By: zdevito
Differential Revision: D9556823
Pulled By: apaszke
fbshipit-source-id: 33e63b17fc7247cac6cfc05eb7eb9bf069b499ee
Summary:
PyTorch exporting test and end to end cases.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/10924
Reviewed By: Ac2zoom
Differential Revision: D9548210
Pulled By: houseroad
fbshipit-source-id: 2381d1ad92a4e07f97060eb65c9fd09f60ad3de6
Summary:
This fixes multiple bugs in the handling of negative indices in both slicing and gather operations. These were uncovered by @[1466077526:Elias Ellison]'s diff D9493614, which made it so that we actually emit negative indices when we see them in PyTorch code.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/10973
Reviewed By: jhcross
Differential Revision: D9546183
Pulled By: jamesr66a
fbshipit-source-id: 6cb0e84e8ad399e47e24a96c44025f644c17b375
Summary:
zdevito et al came to the conclusion that the ONNX spec does not mandate the widening conversion of integral types when serializing tensor data into raw_data, as opposed to serializing the data into int32_data. PyTorch recently made this change in the export code, which caused import in caffe2 to break because it did not match semantics. This fixes that
Pull Request resolved: https://github.com/pytorch/pytorch/pull/10718
Differential Revision: D9423712
Pulled By: jamesr66a
fbshipit-source-id: 479fbae67b028bf4f9c1ca1812c2c7b0c6cccd12
Summary:
Trigonometry functions are newly added to ONNX in a recent PR https://github.com/onnx/onnx/pull/869
This PR makes pytorch support exporting graphs with trigonometry functions.
This PR might need to wait until it is ready to change
```python
_onnx_opset_version = 6
```
to
```python
_onnx_opset_version = 7
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/7540
Differential Revision: D9395041
Pulled By: bddppq
fbshipit-source-id: bdf3e9d212b911c8c4eacf5a0753bb092e4748d2
Summary:
We can't rely on the ATen fallback pathway here because we need to parse out the constant attributes explicitly
Pull Request resolved: https://github.com/pytorch/pytorch/pull/10513
Reviewed By: dzhulgakov
Differential Revision: D9322133
Pulled By: jamesr66a
fbshipit-source-id: 52af947e6c44532ef220cb4b94838ca838b5df06
