Recently, utilizing reinforcement learning (RL) to generate molecules with desired properties has been highlighted as apromising strategy for drug design.

Typicalness-A ware Learning (T AL) to address this issue and improve failure detection performance.

However, in many cases, potential participants in such collaborative schemes are competitors on a downstream task, such as firms that each aim to attract customers by providing the best recommendations.

Wedemonstratetheeffectiveness of our approach for SwAV and ALBERT pretraining in realisticconditions and achieve performance comparable to traditional setups at a fraction of the cost.