Reputation: 69
I'm running a CNN training model with tensorflow. I want to visualize accuracy of the training model with Tensorboard. I tried the following variations:
tensorboard --logdir=/Users/Desktop/model_folder
tensorboard --logdir="/Users/Desktop/model_folder"
But every time I try to load Tensorflow in my browser, the error is: no dashboards are active for the current data set.
I saw some solutions for Windows where they specified the network drive; however, mac does not have a network drive, so I'm not sure how to proceed.
Are there solutions to this error for MacBook? Or are there other ways of visualizing training model accuracy with python?
Upvotes: 0
Views: 123
Reputation: 4357
To simply visualise you don't specifically need tensorboard and can just use history
object that the fit
method returns:
import pandas as pd
#### here you create your model ####
hist = fit(x, y)
histdf = pd.DataFrame(hist)
# Plot loss
hist_df.loc[:, ['train_loss', 'val_loss']].plot(kind='line')
# Plot accuracy
hist_df.loc[:, ['train_accuracy', 'val_accuracy']].plot(kind='line')
You can of course also import matplotlib and set up plots in more detail.
Upvotes: 0