ecobost / cnn4brca

Using Convolutional Neural Networks (CNN) for Semantic Segmentation of Breast Cancer Lesions (BRCA). Master's thesis
75 stars 51 forks source link

error while running ./train.py #6

Closed mohamed945 closed 6 years ago

mohamed945 commented 7 years ago

hello, I'm trying to test your code, but i come with this error,

Traceback (most recent call last): File "train.py", line 174, in train() File "train.py", line 148, in train loss_summary = tf.summary.scalar(['logistic_loss', 'loss'], [train_logistic_loss, train_loss], collections=[]) File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/summary/summary.py", line 118, in scalar name = _clean_tag(name) File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/summary/summary.py", line 90, in _clean_tag new_name = _INVALID_TAGCHARACTERS.sub('', name) TypeError: expected string or bytes-like object

Thanks,

ecobost commented 7 years ago

That line looks like it should be: loss_summary = tf.summary.scalar(['logistic_loss', 'loss'], [train_logistic_loss, train_loss], collections=[])

On Thu, May 11, 2017 at 11:01 AM, mohamed945 notifications@github.com wrote:

hello, I'm trying to test your code, but i come with this error,

File "/train.py", line 148, in train loss_summary = tf.summary.scalar()(['logistic_loss', 'loss'], [train_logistic_loss, train_loss], collections=[]) TypeError: scalar() missing 2 required positional arguments: 'name' and 'tensor'

Thanks,

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/ecobost/cnn4brca/issues/6, or mute the thread https://github.com/notifications/unsubscribe-auth/AJSe5rcJURcDK2coGJxWds5a3_9F-uFzks5r4zDhgaJpZM4NYN6N .

ecobost commented 7 years ago

In tf 1.0,m [summary.scalar] (https://www.tensorflow.org/api_docs/python/tf/summary/scalar), receives only one argument, you would have to modify the code to have each summary sent separately, I bet the new way scalar summaries is managed is different so I would read their docs first. Or you could comment that line entirely out but hen tensorboard won't log the training loss.