Traceback (most recent call last):
File "issue_onnx2torch_002.py", line 15, in
torch_model = convert(onnx_model_path)
File "/opt/conda/envs/tf2onnx/lib/python3.7/site-packages/onnx2torch/converter.py", line 110, in convert
torch_module, onnx_mapping = converter(onnx_node, onnx_graph)
File "/opt/conda/envs/tf2onnx/lib/python3.7/site-packages/onnx2torch/node_converters/averagepool.py", line 48, in
raise NotImplementedError('AvgPool with non symmetrical padding is not implemented.')
NotImplementedError: AvgPool with non symmetrical padding is not implemented.
running the below codes to reproduce:
import numpy as np
import torch
from onnx2torch import convert
Traceback (most recent call last): File "issue_onnx2torch_002.py", line 15, in
torch_model = convert(onnx_model_path)
File "/opt/conda/envs/tf2onnx/lib/python3.7/site-packages/onnx2torch/converter.py", line 110, in convert
torch_module, onnx_mapping = converter(onnx_node, onnx_graph)
File "/opt/conda/envs/tf2onnx/lib/python3.7/site-packages/onnx2torch/node_converters/averagepool.py", line 48, in
raise NotImplementedError('AvgPool with non symmetrical padding is not implemented.')
NotImplementedError: AvgPool with non symmetrical padding is not implemented.
running the below codes to reproduce:
import numpy as np import torch from onnx2torch import convert
def _init_input(input_shape): input_shape = list(input_shape) input_shape[0] = 10 input_shape = tuple(input_shape) input = np.random.rand(*input_shape) return input
onnx_model_path = "../autoconverter/onnx/lenet5-mnist_origin-MDtype0-MDims1-SpecialI5-MParam6-MParam30-MDtype55.onnx" torch_model = convert(onnx_model_path)