jimimvp/torch_rl

Reinforcement learning library for PyTorch.

torch_rl

Reinforcement learning algorithms for spiking networks and artificial neural networks.

Currently implemented

1. Deep deterministic policy gradients with hindsight experience replay
2. Stochastic policy gradient with hindsight experience replay
3. Biased hindsight policy gradient
4. Proximal Policy Optimization on GPU
5. Covariance Matrix Adaptation Evolutionary Strategy

In progress...

1. Distributed proximal policy optimization

torch_rl's Issues

What is the pytorch version?

Hi @jimimvp ,

Thank you for sharing the code!

I am trying to run the ddpg.py script, and run into the following error:

Traceback (most recent call last):
File "examples/ddpg.py", line 7, in
from torch_rl.training import DDPGTrainer
File "/home/z003rjch/workspace/hindsight-pg/torch_rl/training/init.py", line 3, in
from torch_rl.training.ppo import *
File "/home/z003rjch/workspace/hindsight-pg/torch_rl/training/ppo.py", line 371, in
from torch_rl.models.ppo import ActorCriticPPO
File "/home/z003rjch/workspace/hindsight-pg/torch_rl/models/ppo.py", line 4, in
from torch.distributions import Normal, Uniform
ImportError: cannot import name 'Uniform'

I am using pytorch 3.1. I think the error is due to different version of pytorch, so which one are you using?

Thanks!

Best,
Rui