PPOSGD
-
Original paper: https://arxiv.org/abs/1707.06347
-
Baselines blog post: https://blog.openai.com/openai-baselines-ppo/
-
mpirun -np 8 python -m baselines.ppo1.run_atariruns the algorithm for 40M frames = 10M timesteps on an Atari game. See help (-h) for more options. -
python -m baselines.ppo1.run_mujocoruns the algorithm for 1M frames on a Mujoco environment. -
Train mujoco 3d humanoid (with optimal-ish hyperparameters):
mpirun -np 16 python -m baselines.ppo1.run_humanoid --model-path=/path/to/model -
Render the 3d humanoid:
python -m baselines.ppo1.run_humanoid --play --model-path=/path/to/model