In [1]:
import numpy as np
from keras.models import Model
from keras.layers import Input
from keras.layers.convolutional import Conv2DTranspose
from keras import backend as K
import json
from collections import OrderedDict
In [2]:
def format_decimal(arr, places=6):
return [round(x * 10**places) / 10**places for x in arr]
In [3]:
DATA = OrderedDict()
[convolutional.Conv2DTranspose.0] 4 3x3 filters on 4x4x2 input, strides=(1,1), padding='valid', data_format='channels_last', activation='linear', use_bias=False
In [4]:
data_in_shape = (4, 4, 2)
conv = Conv2DTranspose(4, (3,3), strides=(1,1),
padding='valid', data_format='channels_last',
activation='linear', use_bias=False)
layer_0 = Input(shape=data_in_shape)
layer_1 = conv(layer_0)
model = Model(inputs=layer_0, outputs=layer_1)
# set weights to random (use seed for reproducibility)
weights = []
for w in model.get_weights():
np.random.seed(150)
weights.append(2 * np.random.random(w.shape) - 1)
model.set_weights(weights)
print('W shape:', weights[0].shape)
print('W:', format_decimal(weights[0].ravel().tolist()))
# print('b shape:', weights[1].shape)
# print('b:', format_decimal(weights[1].ravel().tolist()))
data_in = 2 * np.random.random(data_in_shape) - 1
result = model.predict(np.array([data_in]))
data_out_shape = result[0].shape
data_in_formatted = format_decimal(data_in.ravel().tolist())
data_out_formatted = format_decimal(result[0].ravel().tolist())
print('')
print('in shape:', data_in_shape)
print('in:', data_in_formatted)
print('out shape:', data_out_shape)
print('out:', data_out_formatted)
DATA['convolutional.Conv2DTranspose.0'] = {
'input': {'data': data_in_formatted, 'shape': data_in_shape},
'weights': [{'data': format_decimal(w.ravel().tolist()), 'shape': w.shape} for w in weights],
'expected': {'data': data_out_formatted, 'shape': data_out_shape}
}
[convolutional.Conv2DTranspose.1] 4 3x3 filters on 4x4x2 input, strides=(1,1), padding='valid', data_format='channels_last', activation='linear', use_bias=True
In [5]:
data_in_shape = (4, 4, 2)
conv = Conv2DTranspose(4, (3,3), strides=(1,1),
padding='valid', data_format='channels_last',
activation='linear', use_bias=True)
layer_0 = Input(shape=data_in_shape)
layer_1 = conv(layer_0)
model = Model(inputs=layer_0, outputs=layer_1)
# set weights to random (use seed for reproducibility)
weights = []
for w in model.get_weights():
np.random.seed(151)
weights.append(2 * np.random.random(w.shape) - 1)
model.set_weights(weights)
print('W shape:', weights[0].shape)
print('W:', format_decimal(weights[0].ravel().tolist()))
print('b shape:', weights[1].shape)
print('b:', format_decimal(weights[1].ravel().tolist()))
data_in = 2 * np.random.random(data_in_shape) - 1
result = model.predict(np.array([data_in]))
data_out_shape = result[0].shape
data_in_formatted = format_decimal(data_in.ravel().tolist())
data_out_formatted = format_decimal(result[0].ravel().tolist())
print('')
print('in shape:', data_in_shape)
print('in:', data_in_formatted)
print('out shape:', data_out_shape)
print('out:', data_out_formatted)
DATA['convolutional.Conv2DTranspose.1'] = {
'input': {'data': data_in_formatted, 'shape': data_in_shape},
'weights': [{'data': format_decimal(w.ravel().tolist()), 'shape': w.shape} for w in weights],
'expected': {'data': data_out_formatted, 'shape': data_out_shape}
}
[convolutional.Conv2DTranspose.2] 4 3x3 filters on 4x4x2 input, strides=(2,2), padding='valid', data_format='channels_last', activation='relu', use_bias=True
In [6]:
data_in_shape = (4, 4, 2)
conv = Conv2DTranspose(4, (3,3), strides=(2,2),
padding='valid', data_format='channels_last',
activation='relu', use_bias=True)
layer_0 = Input(shape=data_in_shape)
layer_1 = conv(layer_0)
model = Model(inputs=layer_0, outputs=layer_1)
# set weights to random (use seed for reproducibility)
weights = []
for w in model.get_weights():
np.random.seed(152)
weights.append(2 * np.random.random(w.shape) - 1)
model.set_weights(weights)
print('W shape:', weights[0].shape)
print('W:', format_decimal(weights[0].ravel().tolist()))
print('b shape:', weights[1].shape)
print('b:', format_decimal(weights[1].ravel().tolist()))
data_in = 2 * np.random.random(data_in_shape) - 1
result = model.predict(np.array([data_in]))
data_out_shape = result[0].shape
data_in_formatted = format_decimal(data_in.ravel().tolist())
data_out_formatted = format_decimal(result[0].ravel().tolist())
print('')
print('in shape:', data_in_shape)
print('in:', data_in_formatted)
print('out shape:', data_out_shape)
print('out:', data_out_formatted)
DATA['convolutional.Conv2DTranspose.2'] = {
'input': {'data': data_in_formatted, 'shape': data_in_shape},
'weights': [{'data': format_decimal(w.ravel().tolist()), 'shape': w.shape} for w in weights],
'expected': {'data': data_out_formatted, 'shape': data_out_shape}
}
[convolutional.Conv2DTranspose.3] 4 3x3 filters on 4x4x2 input, strides=(1,1), padding='same', data_format='channels_last', activation='relu', use_bias=True
In [7]:
data_in_shape = (4, 4, 2)
conv = Conv2DTranspose(4, (3,3), strides=(1,1),
padding='same', data_format='channels_last',
activation='relu', use_bias=True)
layer_0 = Input(shape=data_in_shape)
layer_1 = conv(layer_0)
model = Model(inputs=layer_0, outputs=layer_1)
# set weights to random (use seed for reproducibility)
weights = []
for w in model.get_weights():
np.random.seed(153)
weights.append(2 * np.random.random(w.shape) - 1)
model.set_weights(weights)
print('W shape:', weights[0].shape)
print('W:', format_decimal(weights[0].ravel().tolist()))
print('b shape:', weights[1].shape)
print('b:', format_decimal(weights[1].ravel().tolist()))
data_in = 2 * np.random.random(data_in_shape) - 1
result = model.predict(np.array([data_in]))
data_out_shape = result[0].shape
data_in_formatted = format_decimal(data_in.ravel().tolist())
data_out_formatted = format_decimal(result[0].ravel().tolist())
print('')
print('in shape:', data_in_shape)
print('in:', data_in_formatted)
print('out shape:', data_out_shape)
print('out:', data_out_formatted)
DATA['convolutional.Conv2DTranspose.3'] = {
'input': {'data': data_in_formatted, 'shape': data_in_shape},
'weights': [{'data': format_decimal(w.ravel().tolist()), 'shape': w.shape} for w in weights],
'expected': {'data': data_out_formatted, 'shape': data_out_shape}
}
[convolutional.Conv2DTranspose.4] 5 3x3 filters on 4x4x2 input, strides=(2,2), padding='same', data_format='channels_last', activation='relu', use_bias=True
In [8]:
data_in_shape = (4, 4, 2)
conv = Conv2DTranspose(5, (3,3), strides=(2,2),
padding='same', data_format='channels_last',
activation='relu', use_bias=True)
layer_0 = Input(shape=data_in_shape)
layer_1 = conv(layer_0)
model = Model(inputs=layer_0, outputs=layer_1)
# set weights to random (use seed for reproducibility)
weights = []
for w in model.get_weights():
np.random.seed(154)
weights.append(2 * np.random.random(w.shape) - 1)
model.set_weights(weights)
print('W shape:', weights[0].shape)
print('W:', format_decimal(weights[0].ravel().tolist()))
print('b shape:', weights[1].shape)
print('b:', format_decimal(weights[1].ravel().tolist()))
data_in = 2 * np.random.random(data_in_shape) - 1
result = model.predict(np.array([data_in]))
data_out_shape = result[0].shape
data_in_formatted = format_decimal(data_in.ravel().tolist())
data_out_formatted = format_decimal(result[0].ravel().tolist())
print('')
print('in shape:', data_in_shape)
print('in:', data_in_formatted)
print('out shape:', data_out_shape)
print('out:', data_out_formatted)
DATA['convolutional.Conv2DTranspose.4'] = {
'input': {'data': data_in_formatted, 'shape': data_in_shape},
'weights': [{'data': format_decimal(w.ravel().tolist()), 'shape': w.shape} for w in weights],
'expected': {'data': data_out_formatted, 'shape': data_out_shape}
}
[convolutional.Conv2DTranspose.5] 3 2x3 filters on 4x4x2 input, strides=(1,1), padding='same', data_format='channels_last', activation='relu', use_bias=True
In [9]:
data_in_shape = (4, 4, 2)
conv = Conv2DTranspose(3, (2,3), strides=(1,1),
padding='same', data_format='channels_last',
activation='relu', use_bias=True)
layer_0 = Input(shape=data_in_shape)
layer_1 = conv(layer_0)
model = Model(inputs=layer_0, outputs=layer_1)
# set weights to random (use seed for reproducibility)
weights = []
for w in model.get_weights():
np.random.seed(155)
weights.append(2 * np.random.random(w.shape) - 1)
model.set_weights(weights)
print('W shape:', weights[0].shape)
print('W:', format_decimal(weights[0].ravel().tolist()))
print('b shape:', weights[1].shape)
print('b:', format_decimal(weights[1].ravel().tolist()))
data_in = 2 * np.random.random(data_in_shape) - 1
result = model.predict(np.array([data_in]))
data_out_shape = result[0].shape
data_in_formatted = format_decimal(data_in.ravel().tolist())
data_out_formatted = format_decimal(result[0].ravel().tolist())
print('')
print('in shape:', data_in_shape)
print('in:', data_in_formatted)
print('out shape:', data_out_shape)
print('out:', data_out_formatted)
DATA['convolutional.Conv2DTranspose.5'] = {
'input': {'data': data_in_formatted, 'shape': data_in_shape},
'weights': [{'data': format_decimal(w.ravel().tolist()), 'shape': w.shape} for w in weights],
'expected': {'data': data_out_formatted, 'shape': data_out_shape}
}
In [10]:
print(json.dumps(DATA))
In [ ]: