#! /usr/bin/python
# -*- coding: utf-8 -*-
import numpy as np
import tensorflow as tf
import tensorlayer as tl
from tensorlayer import logging
from tensorlayer.layers.core import Layer, LayerNode
# from tensorlayer.layers.core import LayersConfig
__all__ = ['Input', '_InputLayer']
class _InputLayer(Layer):
"""
The :class:`Input` class is the starting layer of a neural network.
Parameters
----------
shape : tuple (int)
Including batch size.
dtype: dtype
The type of input values. By default, tf.float32.
name : None or str
A unique layer name.
"""
def __init__(self, shape, dtype=tf.float32, name=None): #'input'):
# super(InputLayer, self).__init__(prev_layer=inputs, name=name)
super(_InputLayer, self).__init__(name)
if isinstance(dtype, str):
try:
dtype = eval(dtype)
except Exception as e:
raise RuntimeError("%s is not a valid dtype for InputLayer." % (dtype))
if not isinstance(dtype, tf.DType):
raise RuntimeError("%s is not a valid dtype for InputLayer." % (dtype))
logging.info("Input %s: %s" % (self.name, str(shape)))
self.shape = shape # shape is needed in __repr__
shape_without_none = [_ if _ is not None else 1 for _ in shape]
# self.outputs = self.forward(tl.initializers.random_normal()(shape_without_none))
outputs = self.forward(tl.initializers.ones()(shape_without_none, dtype=dtype))
self._built = True
self._add_node(outputs, outputs)
def __repr__(self):
s = 'Input(shape=%s' % str(self.shape)
if self.name is not None:
s += (', name=\'%s\'' % self.name)
s += ')'
return s
def __call__(self, inputs, *args, **kwargs):
return super(_InputLayer, self).__call__(inputs)
def build(self, inputs_shape):
pass
def forward(self, inputs):
return inputs