Web agents are usually evaluated in live environments, where environment state and judge models drift between runs, so the same checkpoint rarely reproduces the same score, making controlled studies of training phenomena impractical. We present WebMRE, an offline benchmark of 541 tasks and 5,293 steps derived from succ...
Cheng-Guang Gan, Yun-Hao Liang, Qing-Hao Zhang et al.· 0 citations
Instruction-tuning data are judged by quality metrics, and tuned models are judged by benchmarks, but both judgments pass through an output interface: the surface format in which an answer is written. Using gradient signatures across 12 tasks, four semantically equivalent interfaces, three model families, and controlle...
Cheng-Guang Gan, Han-Jun Wei, Yun-Hao Liang et al.· 0 citations
The Mutual Reinforcement Effect is tested in multimodal document understanding on three corpora, two of receipts and one of scanned business forms, comparing single-task, joint and conditioned training, which puts one granularity's gold output in the other's prompt during training only.
Cheng-Guang Gan, Yun-Hao Liang, Han-Jun Wei et al.· 0 citations
This work asks whether it adds skill to a small language and vision-language model web agent at the 4B to 8B scale, or whether it mostly reshapes behavior the supervised model already has, and explains the failure of GRPO.
MAG is introduced, the first benchmark that unifies task execution and guide writing into a single Multimodal Action and Guide task, with two grounding schemes over screenshots: Set-of-Mark element selection and raw pixel coordinates.
