DLR-RM / rl-baselines3-zoo

A training framework for Stable Baselines3 reinforcement learning agents, with hyperparameter optimization and pre-trained agents included.
https://rl-baselines3-zoo.readthedocs.io
MIT License
1.9k stars 495 forks source link

[Question] Plotting Smoothed Evaluation Curve #371

Open Zachary-Fernandes opened 1 year ago

Zachary-Fernandes commented 1 year ago

❓ Question

Hello, I have a question regarding plotting in rl-baselines3-zoo. For plotting the evaluation curves, is there an option to smooth them like one could when plotting the training curves through a rolling window? If not, could one use the same window_func() the training plotter uses or how else would one go about it? I am curious why smoothing is not available. Below is the training curve with a rolling window of 500 and the evaluation curve with no rolling window.

image

image

Checklist

araffin commented 1 year ago

Hello,

s, is there an option to smooth them like one could when plotting the training curves through a rolling window?

There is no option currently, but normally for the evaluation you should have multiple runs and they will smooth out things (this is the all_plot.py script, example is in the doc).

If not, could one use the same window_func() the training plotter uses or how else would one go about it?

you can probably re-use it or see if seaborn has an option for that. (please share your code if you have something working ;))