Skip to content

Latest commit

 

History

History
26 lines (17 loc) · 626 Bytes

README.md

File metadata and controls

26 lines (17 loc) · 626 Bytes

tbplot

Sometimes TensorBoard can take a really long time to load.

tbplot is a little script to plot and save TensorBoard (scalar) graphs more quickly. It'll read logs in parallel using as many cores as you have.

Requirements

  • Python >= 3.6
  • Matplotlib
  • TensorFlow

Usage

$ tbplot --smoothing 0.6 --n_cols 4 run-seed0 run-seed1 run-seed2

(Be careful to supply each run as a separate argument, because tbplot uses its argument list to determine how to separate runs. If you do tbplot ., it'll get confused.)

You'll now have a file run-seed*.png in your current directory: