I am trying to use tensorboard to watch the learning of a convolutional neural net. I am doing good with the tf.summary.merge_all function to create a merged summary. However, I would like to have tracking on accuracy and loss both for training and test data. This post is useful:Logging training and validation loss in tensorboard.
To make things easier to handle, I would like to merge my summaries into two merged summaries, one for training and one for validation.(I will add more stuff eventually, like images weights etc.) I tried to follow the description from tensorboard tf.summary.merge. I can't make it work and I am unable to find any working examples to help me understand where I am going wrong.
with tf.name_scope('accuracy'):
correct_prediction = tf.equal(tf.argmax(y_logits, 1), tf.argmax(y, 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, 'float'))
tf.summary.scalar('accuracy', accuracy)
tf.summary.scalar('train_accuracy', accuracy)
with tf.name_scope('Cost'):
cross_entropy = tf.reduce_mean(
tf.nn.softmax_cross_entropy_with_logits(logits=y_logits, labels=y))
opt = tf.train.AdamOptimizer()
optimizer = opt.minimize(cross_entropy)
grads = opt.compute_gradients(cross_entropy, [b_fc_loc2])
tf.summary.scalar('cost', cross_entropy)
tf.summary.scalar('train_cost', cross_entropy)
with tf.Session() as sess:
writer = tf.summary.FileWriter('./logs/mnistlogs/1f', sess.graph)
sess.run(tf.global_variables_initializer())
merged = tf.summary.merge([cost, accuracy])
This results in the following error:
InvalidArgumentError (see above for traceback): Could not parse one of the summary inputs [[Node: Merge/MergeSummary = MergeSummary[N=2, _device="/job:localhost/replica:0/task:0/cpu:0"](Merge/MergeSummary/inputs_0, Merge/MergeSummary/inputs_1)]]
I would like to know why this doesn't work, and how I can find a solution, any working examples are appreciated.
I figured it out. I need to give the summaries names before merging. The code below solves the problem:
with tf.name_scope('Cost'):
cross_entropy = tf.reduce_mean(
tf.nn.softmax_cross_entropy_with_logits(logits=y_logits, labels=y))
opt = tf.train.AdamOptimizer(learning_rate=0.000003)
optimizer = opt.minimize(cross_entropy)
grads = opt.compute_gradients(cross_entropy, [b_fc_loc2])
cost_sum = tf.summary.scalar('val_cost', cross_entropy)
training_cost_sum = tf.summary.scalar('train_cost', cross_entropy)
with tf.name_scope('accuracy'):
correct_prediction = tf.equal(tf.argmax(y_logits, 1), tf.argmax(y, 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, 'float'))
train_accuracy = accuracy
accuracy_sum = tf.summary.scalar('val_accuracy', accuracy)
training_accuracy_sum = tf.summary.scalar('train_accuracy', accuracy)
with tf.Session() as sess:
writer = tf.summary.FileWriter('./logs/{}/{}'.format(session_name, run_num), sess.graph)
sess.run(tf.global_variables_initializer())
train_merged = tf.summary.merge([training_accuracy_sum, training_cost_sum])
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With