ruslangrimov

import numpy as np
import matplotlib.pyplot as plt
import matplotlib.cm as cm
from itertools import product, count


# generate uniform unit vectors
def generate_grid_vectors(n):
    'Generates matrix NxN of unit length vectors'
    v = np.random.uniform(-1, 1, (n, n, 2))

ruslangrimov

import tensorflow as tf
from tensorflow.python.client import timeline
from keras import backend as K

run_options = tf.RunOptions(trace_level=tf.RunOptions.FULL_TRACE)
run_metadata = tf.RunMetadata()

model = ...  # A Keras model

fn = K.function(model.inputs, model.outputs, options=run_options, run_metadata=run_metadata)

ruslangrimov

from keras import optimizers
from keras import losses
import numpy as np

input_img_data = np.random.random((1,) + K.int_shape(model.inputs[0])[1:])

input_img = K.variable(input_img_data)
inp = Input(tensor=input_img, batch_shape=input_img_data.shape)
out = model(inp)

ruslangrimov

class MyCallBack(Callback):
    def __init__(self, x_v, y_v):
        super(MyCallBack, self).__init__()
        self.x_v, self.y_v = x_v, y_v
        self.ps = []
    
    def on_epoch_end(self, epoch, logs={}):
        p = self.model.predict(self.x_v).flatten()
        self.ps.append(p)
        logs['val_loss'] = log_loss(self.x_y, p)

ruslangrimov

from IPython.display import HTML
import os, base64

def GetLink(fname):
    with open(fname, 'rb') as f:
        data = base64.b64encode(f.read())
        return HTML('<a href="data:application/binary;base64,{0}" download={1}>{1}</a>'.format(data.decode(), os.path.basename(fname)))

ruslangrimov

import tensorflow as tf
from keras.backend.tensorflow_backend import set_session

config = tf.ConfigProto()
config.gpu_options.per_process_gpu_memory_fraction = 0.33
set_session(tf.Session(config=config))

ruslangrimov

from subprocess import Popen, PIPE
import pandas as pd

SYMBOLS = {
    'customary': ('B', 'K', 'M', 'G', 'T', 'P', 'E', 'Z', 'Y'),
    'customary_ext': ('byte', 'kilo', 'mega', 'giga', 'tera', 'peta', 'exa',
                      'zetta', 'iotta'),
    'iec': ('Bi', 'Ki', 'Mi', 'Gi', 'Ti', 'Pi', 'Ei', 'Zi', 'Yi'),
    'iec_60027_2': ('BiB', 'KiB', 'MiB', 'GiB', 'TiB', 'PiB', 'EiB', 'ZiB',
                    'YiB'),

ruslangrimov

'''
Code from https://medium.com/mlreview/a-guide-to-receptive-field-arithmetic-for-convolutional-neural-networks-e0f514068807
'''
import math
# [kernel_size, stride, pad_size], ...
convnet =   [[8, 1, 0], [2, 2, 0],
             [5, 1, 0], [2, 2, 0],
             [3, 1, 0], [2, 2, 0],
             [2, 1, 0], [2, 2, 0]
            ]

ruslangrimov

import os
import keras.backend as K
from keras.models import load_model, model_from_json
import numpy as np
import json


def switch_backend(backend):
    if backend == 'theano':
        # os.environ['KERAS_BACKEND'] = 'theano'

ruslangrimov

import tempfile
from keras.models import load_model
from keras import activations
from keras.layers import Activation

def split_last_layer(model):
    def apply_modifications(model, custom_objects=None):
        model_path = os.path.join(tempfile.gettempdir(), next(tempfile._get_candidate_names()) + '.h5')
        try:
            model.save(model_path)