trying keras to be random seeded (with theano and/or tf)

mnist_nlp_seed.py

'''
trying keras to be random seeded

Switch backends by changin ~/.keras/keras.json:
Theano: https://drive.google.com/open?id=0B-FcStylmYuVU1E5S25QX2JXcGs
TensorFlow: https://drive.google.com/open?id=0B-FcStylmYuVNnpfZERoeUo0QnM 

Keras Backend:
*Theano (1 epoch, GPU): 
Test loss: 0.111780489241
Test accuracy: 0.9667
Time elpased: 0:00:07.026813

*Tensorflow (1 epoch, CPU):
Test loss: 0.797505338073
Test accuracy: 0.7895
Time elpased: 0:00:17.213114

*Tensorflow (1 epoch, CPU):
Test loss: 0.797505338073
Test accuracy: 0.7895
Time elpased: 0:00:17.213114

*Tensorflow (1 epoch, GPU):
DOES NOT SEEM TO BE DETERMINISTIC
Test loss: 0.797505338073
Test accuracy: 0.7895
Time elpased: 0:00:17.213114


Original:
Trains a simple deep NN on the MNIST dataset.

Gets to 98.40% test accuracy after 20 epochs
(there is *a lot* of margin for parameter tuning).
2 seconds per epoch on a K520 GPU.
'''

from __future__ import print_function
import time
from timeit import default_timer as timer
from datetime import timedelta

t_start = timer()

seed = 7

#import os
#os.environ['PYTHONHASHSEED'] = '0' #python 3

import numpy as np
import random as rn
np.random.seed(seed)
rn.seed(seed)

from keras import backend as K
if (K.backend() == 'tensorflow'):
	import tensorflow as tf

	#single thread restriction :(
	session_conf = tf.ConfigProto(
	intra_op_parallelism_threads=1,
	inter_op_parallelism_threads=1)

	from keras import backend as K
	tf.set_random_seed(seed)
	sess = tf.Session(graph=tf.get_default_graph(), config=session_conf)
	#sess = tf.Session(graph=tf.get_default_graph())
	K.set_session(sess)

	#a = tf.random_uniform([1])

	#print("Session 1")
	#with tf.Session() as sess1:
	#	print(sess1.run(a))  # generates 'A1'
	#	print(sess1.run(a))  # generates 'A2'


	#tf.reset_default_graph() #essential when executing inside a notebook

import keras
from keras.datasets import mnist
from keras.models import Sequential
from keras.layers import Dense, Dropout
from keras.optimizers import RMSprop


batch_size = 128
num_classes = 10
epochs = 1

# the data, shuffled and split between train and test sets
(x_train, y_train), (x_test, y_test) = mnist.load_data()

x_train = x_train.reshape(60000, 784)
x_test = x_test.reshape(10000, 784)
x_train = x_train.astype('float32')
x_test = x_test.astype('float32')
x_train /= 255
x_test /= 255
print(x_train.shape[0], 'train samples')
print(x_test.shape[0], 'test samples')

# convert class vectors to binary class matrices
y_train = keras.utils.to_categorical(y_train, num_classes)
y_test = keras.utils.to_categorical(y_test, num_classes)

model = Sequential()
model.add(Dense(512, activation='relu', input_shape=(784,)))
model.add(Dropout(0.2))
model.add(Dense(512, activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(10, activation='softmax'))

model.summary()

model.compile(loss='categorical_crossentropy',
              optimizer=RMSprop(),
              metrics=['accuracy'])

history = model.fit(x_train, y_train,
                    batch_size=batch_size,
                    epochs=epochs,
                    shuffle=True,
                    verbose=0,
                    validation_data=(x_test, y_test))
score = model.evaluate(x_test, y_test, verbose=0)
print();
print('Test loss:', score[0])
print('Test accuracy:', score[1])

print('Time elpased: {}'.format(timedelta(seconds=timer()-t_start)))