Data for Reinforcement Learning Baselines

We provide results for reinforcement learning algorithms.

Results

We provide all results in google drive and baidu drive (extraction code: 0fph).

MuJoCo

• sac-rlkit:

  • code: rlkit with default hyperparameter setting.
  • Algorithm: soft actor critic (SAC). Haarnoja, Tuomas, et al. "Soft actor-critic algorithms and applications." arXiv preprint arXiv:1812.05905 (2018).

• ddpg-rlkit:

  • Code: rlkit with default hyperparameter setting.
  • Algorithm: deep deterministic policy gradient (DDPG). Lillicrap, Timothy P., et al. "Continuous control with deep reinforcement learning." arXiv preprint arXiv:1509.02971 (2015).

• td3:

  • Code: authors' code with default hyperparameter setting.
  • Algorithm: twin delayed deep deterministic policy gradient arglorithm (TD3). Fujimoto, Scott, Herke Van Hoof, and David Meger. "Addressing function approximation error in actor-critic methods." arXiv preprint arXiv:1802.09477 (2018).

• ppo:

  • Code: OpenAI Baselines with default hyperparameter setting.
  • Algorithm: proximal policy optimization (PPO).Schulman, John, et al. "Proximal policy optimization algorithms." arXiv preprint arXiv:1707.06347 (2017).

• trpo:

  • Code: OpenAI Baselines with default hyperparameter setting.
  • Algorithm: trust region policy optimization (TRPO).Schulman, John, et al. "Trust region policy optimization." International conference on machine learning. 2015.

Todo

• More algorithms and implementations, such as the official implementation of sac.
• More than ten random seeds for each algorithm.
• Results for Atari tasks.