In [1]:

    

import numpy as np
import json
from keras.models import Model
from keras.layers import Input
from keras.layers.convolutional import Conv2D
from keras.layers.pooling import MaxPooling2D, AveragePooling2D
from keras.layers.normalization import BatchNormalization
from keras import backend as K
from collections import OrderedDict


    

    




Using TensorFlow backend.

In [2]:

    

def format_decimal(arr, places=6):
    return [round(x * 10**places) / 10**places for x in arr]


    

In [3]:

    

DATA = OrderedDict()


    

In [4]:

    

random_seed = 1011
data_in_shape = (8, 8, 2)

layers = [
    Conv2D(4, (3,3), activation='relu', padding='valid', strides=(1,1), data_format='channels_last', use_bias=True),
    AveragePooling2D(pool_size=(2,2), strides=(1,1), padding='valid', data_format='channels_last')
]

input_layer = Input(shape=data_in_shape)
x = layers[0](input_layer)
for layer in layers[1:-1]:
    x = layer(x)
output_layer = layers[-1](x)
model = Model(inputs=input_layer, outputs=output_layer)

np.random.seed(random_seed)
data_in = 2 * np.random.random(data_in_shape) - 1

# set weights to random (use seed for reproducibility)
weights = []
for i, w in enumerate(model.get_weights()):
    np.random.seed(random_seed + i)
    weights.append(2 * np.random.random(w.shape) - 1)
model.set_weights(weights)

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())

DATA['pipeline_11'] = {
    '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 [5]:

    

import os

filename = '../../test/data/pipeline/11.json'
if not os.path.exists(os.path.dirname(filename)):
    os.makedirs(os.path.dirname(filename))
with open(filename, 'w') as f:
    json.dump(DATA, f)


    

In [6]:

    

print(json.dumps(DATA))


    

    




{"pipeline_11": {"input": {"data": [0.311362, -0.228519, 0.253024, -0.775634, 0.616056, -0.044226, 0.253076, 0.500853, 0.984775, 0.394826, -0.895291, -0.177992, -0.882996, 0.328019, -0.129043, 0.607012, -0.734911, -0.751302, 0.56571, -0.01302, -0.568758, 0.479133, 0.343441, -0.064802, -0.706584, -0.44282, -0.832317, -0.874655, -0.695238, -0.827426, -0.894361, 0.32456, -0.61541, 0.831896, -0.272104, -0.785451, 0.838743, 0.979695, 0.761349, -0.32666, 0.157434, -0.259107, 0.686692, -0.468014, -0.853276, -0.456564, 0.408532, 0.658449, 0.803013, -0.442139, -0.34814, -0.405807, -0.92347, 0.975551, 0.948464, -0.707809, -0.721025, -0.767626, 0.62826, 0.670958, -0.053223, 0.398117, -0.527273, -0.787475, 0.768245, -0.442372, 0.420796, -0.855725, -0.875469, -0.864543, 0.170765, -0.726726, -0.915107, 0.779212, 0.449691, 0.976425, 0.321766, 0.960179, -0.099143, -0.867347, -0.141255, -0.593658, 0.108986, -0.042904, 0.08184, -0.728972, 0.58547, 0.040569, 0.704295, -0.228271, -0.664164, 0.541857, -0.124211, 0.411514, -0.552667, 0.194645, 0.312569, -0.128953, -0.464284, 0.391605, 0.751542, -0.523649, -0.332429, 0.527063, 0.533137, -0.494064, 0.911437, -0.23134, -0.89969, 0.988745, -0.849639, -0.191247, 0.925705, 0.8785, 0.610116, -0.741279, -0.163346, -0.707725, -0.073795, 0.58394, -0.860325, -0.89594, -0.299924, -0.946011, -0.83693, -0.601976, 0.234718, 0.367061], "shape": [8, 8, 2]}, "weights": [{"data": [0.311362, -0.228519, 0.253024, -0.775634, 0.616056, -0.044226, 0.253076, 0.500853, 0.984775, 0.394826, -0.895291, -0.177992, -0.882996, 0.328019, -0.129043, 0.607012, -0.734911, -0.751302, 0.56571, -0.01302, -0.568758, 0.479133, 0.343441, -0.064802, -0.706584, -0.44282, -0.832317, -0.874655, -0.695238, -0.827426, -0.894361, 0.32456, -0.61541, 0.831896, -0.272104, -0.785451, 0.838743, 0.979695, 0.761349, -0.32666, 0.157434, -0.259107, 0.686692, -0.468014, -0.853276, -0.456564, 0.408532, 0.658449, 0.803013, -0.442139, -0.34814, -0.405807, -0.92347, 0.975551, 0.948464, -0.707809, -0.721025, -0.767626, 0.62826, 0.670958, -0.053223, 0.398117, -0.527273, -0.787475, 0.768245, -0.442372, 0.420796, -0.855725, -0.875469, -0.864543, 0.170765, -0.726726], "shape": [3, 3, 2, 4]}, {"data": [-0.946541, 0.585593, -0.49527, 0.594532], "shape": [4]}], "expected": {"data": [0.385866, 1.127311, 0.86104, 0.585367, 0.385866, 1.461206, 0.404148, 0.665295, 1.182823, 0.661164, 0.14916, 0.893525, 1.22361, 0.125944, 0.0, 1.191117, 0.040787, 0.32045, 0.0, 1.735492, 1.208889, 1.18725, 0.254987, 1.936405, 1.208889, 1.604491, 0.254987, 1.498114, 0.419612, 0.417241, 0.0, 1.135943, 0.419612, 0.825821, 0.290601, 0.144114, 0.0, 1.517887, 0.290601, 0.311044, 0.828196, 0.667275, 0.183549, 1.676492, 1.489199, 0.35261, 0.183549, 2.09406, 0.661003, 0.047335, 0.791086, 1.453214, 0.0, 1.311635, 1.081687, 0.225057, 0.0, 2.307613, 0.290601, 0.178278, 1.022356, 0.82228, 0.472916, 0.340965, 1.125167, 0.682943, 0.472916, 1.140136, 1.125167, 0.795228, 1.067056, 0.908882, 0.0, 0.598098, 1.067056, 1.063713, 0.286897, 1.500481, 0.0, 1.379561, 1.076673, 0.620474, 0.464362, 0.336402, 0.573155, 1.000334, 0.289367, 0.321843, 0.690637, 1.814722, 0.27597, 0.165168, 0.117482, 1.158726, 0.27597, 1.595348, 0.548575, 0.982322, 0.0, 2.328448], "shape": [5, 5, 4]}}}

In [ ]: