#! /usr/bin/python
# -*- coding: utf-8 -*-
import tensorlayerx as tlx
from tensorlayerx import logging
from tensorlayerx.nn.core import Module
__all__ = [
'DorefaConv2d',
]
[docs]class DorefaConv2d(Module):
"""The :class:`DorefaConv2d` class is a 2D quantized convolutional layer, which weights are 'bitW' bits and the output of the previous layer
are 'bitA' bits while inferencing.
Note that, the bias vector would not be binarized.
Parameters
----------
bitW : int
The bits of this layer's parameter
bitA : int
The bits of the output of previous layer
out_channels : int
The number of filters.
kernel_size : tuple or int
The filter size (height, width).
stride : tuple or int
The sliding window strides of corresponding input dimensions.
It must be in the same order as the ``shape`` parameter.
act : activation function
The activation function of this layer.
padding : str
The padding algorithm type: "SAME" or "VALID".
data_format : str
"channels_last" (NHWC, default) or "channels_first" (NCHW).
dilation : tuple or int
Specifying the dilation rate to use for dilated convolution.
W_init : initializer or str
The initializer for the the weight matrix.
b_init : initializer or None or str
The initializer for the the bias vector. If None, skip biases.
in_channels : int
The number of in channels.
name : None or str
A unique layer name.
Examples
---------
With TensorLayer
>>> net = tlx.nn.Input([8, 12, 12, 32], name='input')
>>> dorefaconv2d = tlx.nn.DorefaConv2d(
... out_channels=32, kernel_size=(5, 5), stride=(1, 1), act=tlx.ReLU, padding='SAME', name='dorefaconv2d'
... )(net)
>>> print(dorefaconv2d)
>>> output shape : (8, 12, 12, 32)
"""
def __init__(
self,
bitW=1,
bitA=3,
out_channels=32,
kernel_size=(3, 3),
stride=(1, 1),
act=None,
padding='SAME',
data_format="channels_last",
dilation=(1, 1),
W_init='truncated_normal',
b_init='constant',
in_channels=None,
name=None # 'dorefa_cnn2d',
):
super().__init__(name, act=act)
self.bitW = bitW
self.bitA = bitA
self.out_channels = out_channels
self.kernel_size = self.check_param(kernel_size)
self.stride = self._stride = self.check_param(stride)
self.padding = padding
self.data_format = data_format
self.dilation = self._dilation_rate = self.check_param(dilation)
self.W_init = self.str_to_init(W_init)
self.b_init = self.str_to_init(b_init)
self.in_channels = in_channels
if self.in_channels:
self.build(None)
self._built = True
logging.info(
"DorefaConv2d %s: out_channels: %d kernel_size: %s strides: %s pad: %s act: %s" % (
self.name, out_channels, str(kernel_size), str(stride), padding,
self.act.__class__.__name__ if self.act is not None else 'No Activation'
)
)
def __repr__(self):
actstr = self.act.__class__.__name__ if self.act is not None else 'No Activation'
s = (
'{classname}(in_channels={in_channels}, out_channels={out_channels}, kernel_size={kernel_size}'
', strides={strides}, padding={padding}'
)
if self.dilation != (1, ) * len(self.dilation):
s += ', dilation={dilation}'
if self.b_init is None:
s += ', bias=False'
s += (', ' + actstr)
if self.name is not None:
s += ', name=\'{name}\''
s += ')'
return s.format(classname=self.__class__.__name__, **self.__dict__)
def build(self, inputs_shape):
if self.data_format == 'channels_last':
self.data_format = 'NHWC'
if self.in_channels is None:
self.in_channels = inputs_shape[-1]
self._stride = [1, self._stride[0], self._stride[1], 1]
self._dilation_rate = [1, self._dilation_rate[0], self._dilation_rate[1], 1]
elif self.data_format == 'channels_first':
self.data_format = 'NCHW'
if self.in_channels is None:
self.in_channels = inputs_shape[1]
self._stride = [1, 1, self._stride[0], self._stride[1]]
self._dilation_rate = [1, 1, self._dilation_rate[0], self._dilation_rate[1]]
else:
raise Exception("data_format should be either channels_last or channels_first")
self.filter_shape = (self.kernel_size[0], self.kernel_size[1], self.in_channels, self.out_channels)
self.filters = self._get_weights("filters", shape=self.filter_shape, init=self.W_init)
self.b_init_flag = False
if self.b_init:
self.biases = self._get_weights("biases", shape=(self.out_channels, ), init=self.b_init)
self.bias_add = tlx.ops.BiasAdd(self.data_format)
self.b_init_flag = True
self.act_init_flag = False
if self.act:
self.act_init_flag = True
self.dorefaconv2d = tlx.ops.DorefaConv2D(
bitW=self.bitW, bitA=self.bitA, strides=self._stride, padding=self.padding, data_format=self.data_format,
dilations=self._dilation_rate, out_channel=self.out_channels, k_size=self.kernel_size,
in_channel=self.in_channels
)
def forward(self, inputs):
if self._forward_state == False:
if self._built == False:
self.build(tlx.get_tensor_shape(inputs))
self._built = True
self._forward_state = True
outputs = self.dorefaconv2d(inputs, self.filters)
if self.b_init_flag:
outputs = self.bias_add(outputs, self.biases)
if self.act_init_flag:
outputs = self.act(outputs)
if not self._nodes_fixed and self._build_graph:
self._add_node(inputs, outputs)
self._nodes_fixed = True
return outputs