reinforcement learning train_ppo

Official Python bindings with a focus on reinforcement learning and robotics.
Post Reply
vsurobotics2
Posts: 1
Joined: Sun Aug 11, 2019 9:36 pm

reinforcement learning train_ppo

Post by vsurobotics2 » Sun Aug 11, 2019 9:49 pm

hello,

I am trying to run the ppo training example as in the quick start guide
python -m pybullet_envs.agents.train_ppo --config=pybullet_ant --logdir=ant

the train_ppo seems work, but the result policy network does not do anything when tested it with visualize_ppo. anybody have
a working case recently? or there is another forum for this topic?

I am using ubuntu 18, with conda 4.7 and python 3.7. pybullet 2.5.3.

thanks!
J.W

Post Reply