SeungeonBaek / discrete-agents-test

utilizing RL and GNN for trajectory planning(co-work)

Hi!

This repository is for testing RL algorithm with discrete action

(not finished!)

---

Env for algorithm test

lunar-lander
procgen
highway-env

Feature extractor models (..working)

MLPExtractor
RecurrentExtractor
ConvolutionalExtractor (..working)
AttentionExtractor (..working)
GraphExtractor (..working)
CustomExtractor

Single Agent RL Algorithms

DQN family

DQN
DDQN
Agent57 (...working)
REDQ (...working)

distributional DQN

QR_DQN
QUOTA (...working)
IQN
FQF (...working)
MMDQN (...working)
C2D (...working)

others

PPO (...working)
SAC (...working)

RL extension Algorithms

Curiosity-driven RL

ICM
RND
NGU (...working)

for others

MEPPO (...working)
TQC(for SAC) (...working)

---

Multi Agent RL Algorithms

Todo

---

Maintainers

discrete-agents-test is currently maintained by SeungeonBaek
Important Note: We do not do technical support, nor consulting and don't answer personal questions per email.