作者位次 #12/12ACL 2026Findings· Findings of ACL
SOLAR - RL : Semi-Online Long-horizon Assignment Reinforcement Learning
Jichao Wang · Liuyang Bian · Yufeng Zhou · Han Xiao · Yue Pan · Guozhi Wang · Hao Wang · Zhaoxiong Wang · yafei wen · xiaoxin chen · Shuai Ren · Lingfang Zeng
摘要
As Multimodal Large Language Models (MLLMs) mature, GUI agents are evolving from static interactions to complex navigation. While Reinforcement Learning (RL) has emerged as a promising paradigm for training MLLM agents on dynamic GUI tasks, its effective application faces a dilemma.Standard Offli...