ONNX Simplifier

ONNX is great, but sometimes too complicated.

Background

One day I wanted to export the following simple reshape operation to ONNX:

import torch


class JustReshape(torch.nn.Module):
    def __init__(self):
        super(JustReshape, self).__init__()

    def forward(self, x):
        return x.view((x.shape[0], x.shape[1], x.shape[3], x.shape[2]))


net = JustReshape()
model_name = 'just_reshape.onnx'
dummy_input = torch.randn(2, 3, 4, 5)
torch.onnx.export(net, dummy_input, model_name, input_names=['input'], output_names=['output'])

The input shape in this model is static, so what I expected is

However, I got the following complicated model instead:

Our solution

ONNX Simplifier is presented to simplify the ONNX model. It infers the whole computation graph and then replaces the redundant operators with their constant outputs (a.k.a. constant folding).

Features

At its core onnxsim runs a fixed point of shape inference, graph optimization and constant folding until the model stops changing. Around that it offers:

This is the beginning of the README. Read the rest at the original. Read the full README ↗