Skip to content

Commit

Permalink
SAC (#28)
Browse files Browse the repository at this point in the history
  • Loading branch information
yunjhongwu authored Nov 24, 2023
1 parent 2bc40aa commit 3073ffd
Show file tree
Hide file tree
Showing 22 changed files with 4,845 additions and 2,924 deletions.
10 changes: 9 additions & 1 deletion README.md
Original file line number Diff line number Diff line change
Expand Up @@ -4,7 +4,14 @@

![cartpole-training](./assets/cartpole.svg)

**Remark.** This project uses [gym-rs](https://github.com/MathisWellmann/gym-rs) for simulating environments.
### Agents
The project implements the following algorithms:
- Deep Q-Network (DQN)
- Proximal Policy Optimization (PPO)
- Soft Actor-Critic for Discrete Action (SAC-Discrete)

### Environment
This project uses [gym-rs](https://github.com/MathisWellmann/gym-rs) for simulating environments.
Note that the CartPole implementation in gym-rs may lead to slow convergence due to its extended initial state space.
To address this, apply the modification
from [this branch](https://github.com/yunjhongwu/gym-rs/tree/fix-cartpole-starting-state)
Expand All @@ -15,3 +22,4 @@ initialization with the standard behavior of OpenAI Gym CartPole.

- [PyTorch RL tutorial](https://pytorch.org/tutorials/intermediate/reinforcement_q_learning.html)
- [PPO with TorchRL tutorial](https://pytorch.org/rl/tutorials/coding_ppo.html)
- Christodoulou, P. (2019). Soft actor-critic for discrete action settings. *arXiv preprint arXiv:1910.07207*.
Loading

0 comments on commit 3073ffd

Please sign in to comment.