Tensor board
15 likes8,465 views
The document describes how to use TensorBoard, TensorFlow's visualization tool. It outlines 5 steps: 1) annotate nodes in the TensorFlow graph to visualize, 2) merge summaries, 3) create a writer, 4) run the merged summary and write it, 5) launch TensorBoard pointing to the log directory. TensorBoard can visualize the TensorFlow graph, plot metrics over time, and show additional data like histograms and scalars.
Tensor board
- 1. Visualizing your Deep learning using TensorBoard (TensorFlow) Sung Kim <hunkim+ml@gmail.com>
- 2. TensorBoard:TF logging/debugging tool • Visualize your TF graph • Plot quantitative metrics • Show additional data https://www.tensorflow.org/versions/r0.7/how_tos/summaries_and_tensorboard/index.html
- 3. Old fashion: print, print, print
- 4. New way!
- 5. 5 steps of using tensorboard • From TF graph, decide which node you want to annotate - with tf.name_scope("test") as scope: - tf.histogram_summary("weights",W), tf.scalar_summary(“accuracy", accuracy) • Merge all summaries - merged = tf.merge_all_summaries() • Create writer - writer = tf.train.SummaryWriter("/tmp/mnist_logs", sess.graph_def) • Run summary merge and add_summary - summary = sess.run(merged, …); writer.add_summary(summary); • Launch Tensorboard - tensorboard --logdir=/tmp/mnist_logs
- 7. Add scope for better graph hierarch
- 10. Merge summaries and create writer after creating session
- 11. Run merged summary and write (add summary)
- 12. Launch tensorboard • tensorboard —logdir=/tmp/mnist_logs • (You can navigate to http://0.0.0.0:6006)
- 13. Add scope for better graph hierarch
- 14. Add scope for better graph hierarch
- 15. Add histogram
- 17. 5 steps of using tensorboard • From TF graph, decide which node you want to annotate - with tf.name_scope("test") as scope: - tf.histogram_summary("weights",W), tf.scalar_summary(“accuracy", accuracy) • Merge all summaries - merged = tf.merge_all_summaries() • Create writer - writer = tf.train.SummaryWriter("/tmp/mnist_logs", sess.graph_def) • Run summary merge and add_summary - summary = sess.run(merged, …); writer.add_summary(summary); • Launch Tensorboard - tensorboard --logdir=/tmp/mnist_logs