You are viewing a single comment's thread from:

RE: LeoThread 2024-11-08 06:46

in LeoFinance2 months ago

Tsinghua University and Zhipu AI researchers created WEBRL to train web agents using open LLMs, helping them learn by trial and error. It means that, instead of preset tasks, agents evolve by solving new tasks generated from past mistakes. This new approach shows promise for smarter, autonomous web agents.

#AI #ReinforcementLearning