CODEWITHSUNDEEP
pythontensorflowdeep-learningtensorboard
soshi shimada
soshi shimada

Reputation: 435

Tensor board doesn't show scalars

I wrote mnist code using Multi Layer Perceptrons. But it doesn't show scalars of accuracy and loss function.(but it successfully shows a graph of a model) If you know, could you give me a clue? Tensorflow version:1.2.0

These are the functions which I want to show in Tensorboard.

def loss(label,y_inf):
    # Cost Function basic term
    with tf.name_scope('loss'):
        cross_entropy = -tf.reduce_sum(label * tf.log(y_inf))
    tf.summary.scalar("cross_entropy", cross_entropy)
    return cross_entropy



def accuracy(y_inf, labels):
    with tf.name_scope('accuracy'):
        correct_prediction = tf.equal(tf.argmax(y_inf, 1), tf.argmax(labels, 1))
        accuracy = tf.reduce_mean(tf.cast(correct_prediction, "float"))
    tf.summary.scalar("accuracy", accuracy)
    return accuracy

Upvotes: 1

Views: 1357

Answers (1)

amirbar
amirbar

Reputation: 839

One thing you might be missing is to actually fetch these summaries and write them to disk.

First, you have to define a FileWriter:

fw = tf.summary.FileWriter(LOGS_DIR) # LOGS_DIR should correspond to the path you want to save the summaries in

Next, merge all your summaries into a single op:

summaries_op = tf.summary.merge_all()

Now, within your training loop, make sure you write the summaries to disk:

for i in range(NUM_ITR):
    _, summaries_str = sess.run([train_op, summaries_op])
    fw.add_summary(summaries_str, global_step=i)

In order to see these summaries in tensorboard run:

tensorboard --logdir=LOGS_DIR

Upvotes: 1

Related Questions