Could you please add the recently released ARPO reinforcement learning algorithm?

[EngineerChao]

The ARPO algorithm effectively improves the performance of multi-round tool inference agents and solves the problems of insufficient exploration and lack of generalization ability of existing sample-level RL methods in multi-round interactions. github: https://github.com/dongguanting/ARPO

[lunaqiu]

@EngineerChao Thank you for your suggestion! This looks like a very interesting approach that could indeed improve the scenarios you listed. Since we are severely short of hands, would you be interested in collaborating with us to implement this feature? Your expertise would be valuable in integrating ARPO effectively.