mirror of
https://github.com/pytorch/pytorch.git
synced 2025-10-20 21:14:14 +08:00
9bbe1d632e
This PR introduces 3 BC changes: First, this PR propagates `BUILD_CAFFE2` flag to `libtorch` and `libtorch_python`, which is necessary for non-caffe2 ONNX runtimes when using `ONNX_ATEN_FALLBACK` operator export type. Second, as a complement of https://github.com/pytorch/pytorch/pull/68490, this PR refactors Caffe2's Aten ops symbolics to consider not only the `operator_export_type` (aka `ONNX_ATEN_FALLBACK`) to emit Caffe2 Aten ops, but also whether `BUILD_CAFFE2` (which is called `torch.onnx._CAFFE2_ATEN_FALLBACK` in python binding) is set. Lastly, it renames `onnx::ATen` to `aten::ATen` for ONNX spec consistency in a BC fashion. ONNX doesn't have `ATen` op on its spec, but PyTorch ONNX converter emits them. Non-Caffe2 backend engines would be mislead by such operator's name/domain. A non-ideal workaround would be to have Aten ops handled based on its name and ignore the (non-complaint) domain. Moreover, users could incorrectly file bugs to either ONNX or ONNX Runtime when they inspect the model and notice the presence of an unspecified ONNX operator. Pull Request resolved: https://github.com/pytorch/pytorch/pull/73954 Approved by: https://github.com/BowenBao, https://github.com/malfet, https://github.com/garymm, https://github.com/jiafatom
21 lines
999 B
Plaintext
21 lines
999 B
Plaintext
ModelProto {
|
|
producer_name: "pytorch"
|
|
domain: ""
|
|
doc_string: ""
|
|
graph:
|
|
GraphProto {
|
|
name: "torch-jit-export"
|
|
inputs: [{name: "0", type:Tensor dims: 3 4}]
|
|
outputs: [{name: "5", type:Tensor dims: 0}]
|
|
initializers: []
|
|
nodes: [
|
|
Node {type: "Constant", inputs: [], outputs: [1], attributes: [{ name: 'value', type: tensor, value:TensorProto shape: []}]},
|
|
Node {type: "Less", inputs: [0,1], outputs: [2], attributes: []},
|
|
Node {type: "Cast", inputs: [2], outputs: [3], attributes: [{ name: 'to', type: int, value: 2}]},
|
|
Node {type: "Cast", inputs: [3], outputs: [4], attributes: [{ name: 'to', type: int, value: 9}]},
|
|
Node {type: "ATen", domain: "org.pytorch.aten", inputs: [0,4], outputs: [5], attributes: [{ name: 'operator', type: string, value: 'index'}, { name: 'overload_name', type: string, value: ''}]}
|
|
]
|
|
}
|
|
opset_import: [OperatorSetIdProto { domain: }OperatorSetIdProto { domain: org.pytorch.aten}],
|
|
}
|