for k in range(folds):
print("Fold: {0}".format(k))
trainingData = data[0][indices[k][0]]
trainingLabels = data[1][indices[k][0]]
testingData = data[0][indices[k][1]]
testingLabels = data[1][indices[k][1]]
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
# Training
for i in range(epochs):
# Batches
for j in range(0, trainingData.shape[0], b_size):
x_raw = trainingData[j:j + b_size]
y_raw = hotvector(trainingLabels[j:j + b_size], classes)
[la, c] = sess.run([optimizer, cost],
feed_dict={
x: x_raw,
y: y_raw,
phase: True
})
#saver.save(sess, 'tmp/my-weights')
#g = sess.graph
#gdef = g.as_graph_def()
#tf.train.write_graph(gdef,"tmp","graph.pb",False)
# Testing
c = 0
print(counter,len(allConfgs))
(x,y),optimizer,cost,eval_pred,saver = getModel(conf)
b_size = conf[-1]
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
merged = tf.summary.merge_all()
writer = tf.summary.FileWriter("logs/"+conf2str(conf),sess.graph)
step = 0
# Training
for i in range(epochs):
# Batches
for j in range(0,data["train"][1].shape[0],b_size):
x_raw = data["train"][0][j:j+b_size]
y_raw = hotvector(data["train"][1][j:j+b_size],classes)
[la,c,summary]=sess.run([optimizer,cost,merged], feed_dict={x: x_raw, y: y_raw})
writer.add_summary(summary,step)
step+=1
writer.close()
#saver.save(sess, 'tmp/my-weights')
#g = sess.graph
#gdef = g.as_graph_def()
#tf.train.write_graph(gdef,"tmp","graph.pb",False)
# Testing
c=0;g=0
goodones=[]