search

araffin / rl-baselines-zoo

A collection of 100+ pre-trained RL agents using Stable Baselines, training and hyperparameter optimization included.
https://stable-baselines.readthedocs.io/
MIT License
1.12k stars 206 forks source link

Using the monitor wrapper with ppo2 #95

Closed meric-sakarya closed 4 years ago

meric-sakarya commented 4 years ago

I am trying to use the Monitor Wrapper to better visualize the results of my training. I added the following line to the ppo2.yml file: env_wrapper: stable_baselines.bench.monitor. It did not work and I received this error when trying to restart the training: 2020-07-25 (1)

I also tried to copy the file Monitor from stable baselines to the wrappers.py file in my Zoo folder and use that directly but that also did not work: 2020-07-25 (2)

araffin commented 4 years ago

If you use the zoo, the env is automatically wrapped with a Monitor wrapper.

Please fill the issue template properly next time (using markdown codeblock).

meric-sakarya commented 4 years ago

The environment uses more than one environment and therefore the graph results on Tensorboard are wacky. Just like in this issue hill-a/stable-baselines#143. 2020-07-25

araffin commented 4 years ago

The environment uses more than one environment and therefore the graph results on Tensorboard are wacky. Just like in this issue hill-a/stable-baselines#143.

This is unrelated to the original question. You should take a look at the documentation first to plot the information displayed in the terminal: https://stable-baselines.readthedocs.io/en/master/guide/tensorboard.html#legacy-integration