a2rl_pytorch

This is an unofficial pytorch implementation of A2-RL[1, 2]. In this implementation, we provide the training procedure. Note that this implementation is based on A2C(Actor-Critic).

Reference

[1] A2-RL: Aesthetics Aware Reinforcement Learning for Image Cropping Debang

[2] https://github.com/wuhuikai/TF-A2RL