mingxiaoh / onnx-chainer

Add-on package for ONNX format support in Chainer

ONNX-Chainer

This is an add-on package for ONNX support by Chainer.

Requirements

• onnx==0.2.1
• chainer>=3.2.0

Installation

See INSTALL.md

Quick Start

import numpy as np
import chainer
import chainer.links as L
import onnx_chainer

model = L.VGG16Layers()

# Pseudo input
x = np.zeros((1, 3, 224, 224), dtype=np.float32)

# Don't forget to set train flag off!
chainer.config.train = False

onnx_chainer.export(model, x, filename='VGG16.onnx')

Load models from Caffe2

Using onnx-caffe2 is a simple way to do it.

import chainer
import chainer.links as L
import numpy as np

from onnx_caffe2.backend import Caffe2Backend
from onnx_caffe2.backend import run_model
from onnx_caffe2.helper import save_caffe2_net

import onnx_chainer

# Instantiate a Chainer model (Chain object)
model = L.VGG16Layers()

# Prepare a dummy input
x = np.random.randn(1, 3, 224, 224).astype(np.float32)

# Do not forget setting train flag off!
chainer.config.train = False

# Export to ONNX model
onnx_model = onnx_chainer.export(model, x)

# Convert ONNX model to Caffe2 model
init_net, predict_net = Caffe2Backend.onnx_graph_to_caffe2_net(
    onnx_model.graph, device='CPU')

# Save the Caffe2 model to disk
init_file = "./vgg16_init.pb"
predict_file = "./vgg16_predict.pb"
save_caffe2_net(init_net, init_file, output_txt=False)
save_caffe2_net(predict_net, predict_file, output_txt=True)

# Run the model with Caffe2
caffe2_out = run_model(onnx_model, [x])[0]

Supported Functions

Currently 50 Chainer Functions are supported to export in ONNX format.

Activation

• ELU
• HardSigmoid
• LeakyReLU
• LogSoftmax
• PReLUFunction
• ReLU
• Sigmoid
• Softmax
• Softplus
• Tanh

Array

• Cast
• Concat
• Depth2Space
• Pad ¹²
• Reshape
• Space2Depth
• SplitAxis
• Squeeze
• Tile
• Transpose

Connection

• Convolution2DFunction
• ConvolutionND
• Deconvolution2DFunction
• DeconvolutionND
• EmbedIDFunction ³
• LinearFunction

Math

• Add
• Absolute
• Div
• Mul
• Neg
• PowVarConst
• Sub
• Clip
• Exp
• Identity
• MatMul ⁴
• Maximum
• Minimum
• Sqrt
• SquaredDifference
• Sum

Noise

• Dropout ⁵

Pooling

• AveragePooling2D
• AveragePoolingND
• MaxPooling2D
• MaxPoolingND

Normalization

• BatchNormalization
• FixedBatchNormalization
• LocalResponseNormalization

---

1: mode should be either 'constant', 'reflect', or 'edge'
2: ONNX doesn't support multiple constant values for Pad operation
3: Current ONNX doesn't support ignore_label for EmbedID
4: Current ONNX doesn't support transpose options for matmul ops
5: In test mode, all dropout layers aren't included in the exported file