In [1]:

    

import numpy as np
import json
from keras.models import Model
from keras.layers import Input, Conv2D, SeparableConv2D, Activation, BatchNormalization
from keras import backend as K
from collections import OrderedDict


    

    




Using TensorFlow backend.
/home/leon/miniconda3/lib/python3.6/importlib/_bootstrap.py:219: RuntimeWarning: compiletime version 3.5 of module 'tensorflow.python.framework.fast_tensor_util' does not match runtime version 3.6
  return f(*args, **kwds)

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 = 10015
data_in_shape = (7, 7, 1)

layers = [
    Conv2D(2, (3,3), padding='valid', strides=(1,1)),
    BatchNormalization(epsilon=1e-03, axis=-1, center=True, scale=True),
    Activation('relu'),
    SeparableConv2D(2, (3,3), padding='same', strides=(1,1)),
    BatchNormalization(epsilon=1e-03, axis=-1, center=True, scale=True)
]

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)
    if i == 5 or i == 12:
        # std should be positive
        weights.append(np.random.random(w.shape))
    else:
        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_15'] = {
    '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/15.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_15": {"input": {"data": [0.263717, -0.720003, 0.8268, -0.92253, 0.462425, 0.244446, -0.124898, -0.811324, 0.354873, 0.801179, 0.564291, -0.355745, 0.644611, 0.859794, -0.560491, -0.788843, -0.963575, -0.638994, 0.405345, 0.176622, -0.153501, 0.998311, 0.944831, 0.536878, 0.033269, -0.435141, -0.309305, 0.602218, 0.537614, 0.254239, 0.37362, -0.505583, -0.948316, 0.939032, 0.149339, -0.197973, -0.224213, 0.028595, -0.795155, -0.951132, 0.729428, 0.59133, 0.864157, -0.620287, -0.843726, -0.920702, -0.496038, 0.822909, 0.890635], "shape": [7, 7, 1]}, "weights": [{"data": [0.263717, -0.720003, 0.8268, -0.92253, 0.462425, 0.244446, -0.124898, -0.811324, 0.354873, 0.801179, 0.564291, -0.355745, 0.644611, 0.859794, -0.560491, -0.788843, -0.963575, -0.638994], "shape": [3, 3, 1, 2]}, {"data": [-0.296519, 0.572188], "shape": [2]}, {"data": [-0.983721, 0.040472], "shape": [2]}, {"data": [0.524698, 0.474407], "shape": [2]}, {"data": [-0.040091, -0.549528], "shape": [2]}, {"data": [0.14783, 0.198852], "shape": [2]}, {"data": [0.584586, -0.737099, -0.575981, -0.959165, 0.528879, -0.767619, -0.669488, -0.008686, 0.12889, -0.992103, 0.481899, -0.20261, 0.81531, 0.405667, -0.500145, -0.349009, 0.426971, 0.662575], "shape": [3, 3, 2, 1]}, {"data": [-0.470443, -0.553947, -0.078266, 0.730533], "shape": [1, 1, 2, 2]}, {"data": [-0.389418, 0.767296], "shape": [2]}, {"data": [0.550792, -0.229108], "shape": [2]}, {"data": [0.304147, -0.117505], "shape": [2]}, {"data": [0.453662, -0.506543], "shape": [2]}, {"data": [0.296361, 0.407461], "shape": [2]}], "expected": {"data": [0.219238, -0.677985, -2.58341, 0.360723, -1.275605, -0.228328, 0.392925, -0.896873, 0.417156, -0.834217, -0.415627, -0.135939, -1.048117, 0.123814, -1.744815, 0.405, -1.046245, 0.134385, -1.565544, 0.23616, -0.536067, -0.116145, -0.326501, -0.122216, -0.831852, 0.144915, -1.604822, 0.340508, 0.254569, -0.499731, -0.276547, -0.252957, -1.31593, 0.320422, -2.877554, 0.97416, -2.366373, 0.673242, -0.941472, -0.02324, -0.42089, -0.15502, -1.318147, 0.341634, -2.617689, 0.895466, 0.085786, -0.218687, 1.115638, -0.832125], "shape": [5, 5, 2]}}}

In [ ]: