DocsTensorboard
Experiment TrackingVisualizationsTensorboard

Overview

Tensorboard is a tool that is used by data scientists to visualize and debug deep learning models. Polyaxon provides several ways for using tensorboard:

  • Tensorboard for a single run
  • Tensorboard for a multi-run
  • Tensorboard based on metric(s) performance or a search

Single run

This TensorBoard service loads only the tuned classifier run. Its scalar charts and weight histograms show data from that single run.

run-dashboards-tensorboard

run-dashboards-tensorboard-histo

Comparing runs

A separate TensorBoard service loads three runs: baseline, compact, and tuned. Select all three to compare their curves, or expand a chart for a closer view.

comparison-tensorboard-compare

Programmatic Experience
Runs dashboard

On this page

OverviewSingle runComparing runs

Actions

Give us feedbackEdit this page on GitHub