Is there a script for loading and visualizing trained policies?

I have gathered that train.py saves the actor_critic at a specified interval. Is there an easy way to load, run, and visualize the saved actor_critic?

If you mean to just load it and visualize it in the standard GUI, then you should be able to do it using the --test flag like this (the method is different if you are using the rl_games package):

python train.py --task=FrankaCabinet --num_envs=1 --test --resume=1000 --seed=1

Using the --resume flag should load the weights that were saved after 1000 iterations and is stored in your python/rlgpu/logs/ directory (in this case there should be a FrankaCabinet directory full of trained models.

If you have used the --logdir flag to save the weights elsewhere you would want to use it again here to specify where the code should find your stored weights.

If you want an overview of the available flags, check the python/rlgpu/utils/config.py script or run python train.py --help.

1 Like

Perfect, thanks so much.

1 Like

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.