- Original paper: https://arxiv.org/abs/1502.05477
- Baselines blog post https://blog.openai.com/openai-baselines-ppo/
mpirun -np 16 python -m baselines.trpo_mpi.run_atariruns the algorithm for 40M frames = 10M timesteps on an Atari game. See help (-h) for more options.python -m baselines.trpo_mpi.run_mujocoruns the algorithm for 1M timesteps on a Mujoco environment.
trpo_mpi
Folders and files
| Name | Name | Last commit date | ||
|---|---|---|---|---|
parent directory.. | ||||