onnx2torch
onnx2torch is an ONNX to PyTorch converter. Our converter:
- Is easy to use – Convert the ONNX model with the function call
convert; - Is easy to extend – Write your own custom layer in PyTorch and register it with
@add_converter; - Convert back to ONNX – You can convert the model back to ONNX using the
torch.onnx.exportfunction.
If you find an issue, please let us know! And feel free to create merge requests.
Please note that this converter covers only a limited number of PyTorch / ONNX models and operations.
Let us know which models you use or want to convert from onnx to torch here.
Installation
From PyPi
pip install onnx2torch
Usage
Below you can find some examples of use.
Convert
import torch
from onnx2torch.converter import convert
# Path to ONNX model
onnx_model_path = '/some/path/mobile_net_v2.onnx'
# You can pass the path to the onnx model to convert it or...
torch_model_1 = convert(onnx_model_path)
# Or you can load a regular onnx model and pass it to the converter
onnx_model = onnx.load(onnx_model_path)
torch_model_2 = convert(onnx_model)
Execute
We can execute the returned PyTorch model in the same way as the original torch model.
import onnxruntime as ort
# Create example data
x = torch.ones((1, 2, 224, 224)).cuda()
out_torch = torch_model_1(x)
ort_sess = ort.InferenceSession(onnx_model_path)
outputs_ort = ort_sess.run(None, {'input': x.numpy()})
# Check the Onnx output against PyTorch
print(torch.max(torch.abs(outputs_ort - out_torch.detach().numpy())))
print(np.allclose(outputs_ort, out_torch.detach().numpy(), atol=1.e-7))
Models
We have tested the following models:
- ResNet50
- SSDLite with MobileNetV2 backbone
How to add new operations to converter
Here we show how to add the module:
- Supported by both PyTorch and ONNX and has the same behaviour.
An example of such a module is Relu
@add_converter(operation_type='Relu', version=6)
@add_converter(operation_type='Relu', version=13)
@add_converter(operation_type='Relu', version=14)
def _(node: OnnxNode, graph: OnnxGraph) -> OperationConverterResult:
return OperationConverterResult(
torch_module=nn.ReLU(),
onnx_mapping=onnx_mapping_from_node(node=node),
)
Here we have registered an operation named Relu for opset versions 6, 13, 14.
Note that the torch_module argument in OperationConverterResult must be a torch.nn.Module, not just a callable object!
If Operation's behaviour differs from one opset version to another, you should implement it separately.
- Operations supported by PyTorch and ONNX BUT have different behaviour
class OnnxExpand(nn.Module):
@staticmethod
def _do_forward(input_tensor: torch.Tensor, shape: torch.Tensor) -> torch.Tensor:
return input_tensor * torch.ones(torch.Size(shape), dtype=input_tensor.dtype, device=input_tensor.device)
def forward(self, *args) -> torch.Tensor:
if torch.onnx.is_in_onnx_export():
with skip_torch_tracing():
output = self._do_forward(*args)
return _ExpandExportToOnnx.set_output_and_apply(output, *args)
return self._do_forward(*args)
class _ExpandExportToOnnx(CustomExportToOnnx):
@staticmethod
def symbolic(graph: torch_C.Graph, *args, **kwargs) -> torch_C.Value:
return graph.op('Expand', *args, **kwargs, outputs=1)
@add_converter(operation_type='Expand', version=8)
@add_converter(operation_type='Expand', version=13)
def _(node: OnnxNode, graph: OnnxGraph) -> OperationConverterResult: # pylint: disable=unused-argument
return OperationConverterResult(
torch_module=OnnxExpand(),
onnx_mapping=onnx_mapping_from_node(node=node),
)
Here we have used a trick to convert the model from torch back to ONNX by defining the custom _ExpandExportToOnnx.
![[Code Quality] Remove trailing whitespace and add isort for module import](https://avatars.githubusercontent.com/u/92794867?v=4)
and its forward
6 May 15, 2022
5.7k Jan 9, 2023
264 Dec 30, 2022
18 Nov 6, 2022
1.8k Jan 8, 2023
5 Nov 19, 2022
1.5k Dec 31, 2022
11 Mar 20, 2022
2 Feb 16, 2022
4.2k Jan 1, 2023
34.1k Dec 31, 2022
94 Jan 6, 2023