Visualizing and Comparing Multiple OpenAI Gym Experiments
Me too.
- Start with the notebook, see if it fits your needs.
- Copy over some experimenal results worth comparing, there are currently a few example folders here so that you can get a sense of the structure and funcionality.
- Revel in the beautiful aggregate statistics.
Currently, gym-multiplot displays average and standard deviation for episode rewards and lengths. Figures show mean in black and each run in a different slightly transparent color, linked to the legend. Standard deviation is the shaded gray region around the mean.
First, this was a proof of concept hackup. Second, drop an issue and let's figure out how to make this work more robustly together.
There are lots more statistics to capture from OpenAI gym, and this just shows a single sample...
JKCooper2 for the quick help early on.