gouxiangchen/soft-Q-learning

discrete soft Q learning(SQL) and soft Q imitation learning(SQIL) implementation in pytorch, simple!

SQL & SQIL

the implement of soft Q learning algorithm in pytorch

note that this is for discrete action space

update SQIL: soft q imitation learning

all code is in one file and easily to follow

requirment

• tensorboardX (for logging, you can delete the logging code if you don't need)
• pytorch (>= 1.0, 1.0.1 used in my experiment)
• gym

in Cartpole-v0

Ref

Reinforcement Learning with Deep Energy-Based Policies

SQIL: Imitation Learning via Reinforcement Learning with Sparse Rewards