In [1]:
import numpy as np
from keras.models import Model
from keras.layers import Input
from keras.layers.convolutional import Conv3D
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.Conv3D.0] 4 3x3x3 filters on 5x5x5x2 input, strides=(1,1,1), padding='valid', data_format='channels_last', dilation_rate=(1,1,1), activation='linear', use_bias=True
In [4]:
data_in_shape = (5, 5, 5, 2)
conv = Conv3D(4, (3,3,3), strides=(1,1,1), padding='valid',
data_format='channels_last', dilation_rate=(1,1,1),
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(130)
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.Conv3D.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.Conv3D.1] 2 3x3x3 filters on 4x4x4x2 input, strides=(1,1,1), padding='valid', data_format='channels_last', dilation_rate=(1,1,1), activation='sigmoid', use_bias=False
In [5]:
data_in_shape = (4, 4, 4, 2)
conv = Conv3D(2, (3,3,3), strides=(1,1,1), padding='valid',
data_format='channels_last', dilation_rate=(1,1,1),
activation='sigmoid', 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(131)
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.Conv3D.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.Conv3D.2] 2 3x3x3 filters on 4x4x3x2 input, strides=(1,1,1), padding='same', data_format='channels_last', dilation_rate=(1,1,1), activation='relu', use_bias=True
In [6]:
data_in_shape = (4, 4, 3, 2)
conv = Conv3D(2, (3,3,3), strides=(1,1,1), padding='same',
data_format='channels_last', dilation_rate=(1,1,1),
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(132)
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.Conv3D.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.Conv3D.3] 2 3x3x2 filters on 4x4x3x2 input, strides=(2,1,1), padding='same', data_format='channels_last', dilation_rate=(1,1,1), activation='relu', use_bias=True
In [7]:
data_in_shape = (4, 4, 3, 2)
conv = Conv3D(2, (3,3,2), strides=(2,1,1), padding='same',
data_format='channels_last', dilation_rate=(1,1,1),
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(133)
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.Conv3D.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.Conv3D.4] 2 3x3x3 filters on 6x6x4x2 input, strides=(3,3,2), padding='same', data_format='channels_last', dilation_rate=(1,1,1), activation='relu', use_bias=True
In [8]:
data_in_shape = (6, 6, 4, 2)
conv = Conv3D(2, (3,3,3), strides=(3,3,2), padding='same',
data_format='channels_last', dilation_rate=(1,1,1),
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(134)
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.Conv3D.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.Conv3D.5] 2 3x3x3 filters on 6x4x4x2 input, strides=(1,1,1), padding='valid', data_format='channels_last', dilation_rate=(2,1,1), activation='relu', use_bias=True
In [9]:
data_in_shape = (6, 4, 4, 2)
conv = Conv3D(2, (3,3,3), strides=(1,1,1), padding='valid',
data_format='channels_last', dilation_rate=(2,1,1),
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(135)
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.Conv3D.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}
}
[convolutional.Conv3D.6] 2 3x3x3 filters on 4x4x3x2 input, strides=(1,1,1), padding='same', data_format='channels_last', dilation_rate=(1,2,1), activation='relu', use_bias=True
In [10]:
data_in_shape = (4, 4, 3, 2)
conv = Conv3D(2, (3,3,3), strides=(1,1,1), padding='same',
data_format='channels_last', dilation_rate=(1,2,1),
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(136)
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.Conv3D.6'] = {
'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 [11]:
print(json.dumps(DATA))