Contrastive Learning as Goal-Conditioned Reinforcement Learning

Did really good in some of the OGbench tasks as shown in the Horizon Reduction Makes RL Scalable.

Citation

(Eysenbach et al., 2022)

Eysenbach, B., Zhang, T., Salakhutdinov, R., & Levine, S. (2022). Contrastive Learning as Goal-Conditioned Reinforcement Learning. https://arxiv.org/abs/2206.07568