robotics-rl-srl
robotics-rl-srl copied to clipboard
araffin
S-RL Toolbox: Reinforcement Learning (RL) and State Representation Learning (SRL) for Robotics
## Problem description In the current (origin/master) version, there are two preprocessing mode: 1. 'tf' and 2. 'image_net'. I have noticed that with the some device/python environment/encoder model/robot environment etc,...
**Describe the bug** The training of SRL algorithm slows down by a few seconds per epoch. The first epoch cost 31s then, it grows almost linearly to 60s at the...
**Describe the bug** In the scripts rl_baselines/rl_algorithm/sac.py, deepq.py, ddpg.py: def customArguments(self, parser): ## forget to add: super().customArguments(parser) **Code example** The following command line reproduces the issue: - python -m rl_baselines.train...
**Describe the bug** - [1.] ValueError: Seed must be between 0 and 2**32 - 1 - [2.] ImportError: cannot import name 'prng' (for gym>=0.11.0) **Code example** The following code can...
I have noticed that the images generated by the environment are inconsistent with the other features. In fact, due to the 'channel first' convention in Pytorch, the images should be...
When you use the first way to add your SRL model, cf README of repo state_representation : > If your SRL model is a carateristic of the environment (position, angles,...
