I-Hung Hsu
Titles as the sources gave them. In 1 entry, with 0 checked quotations.
In the Base
Every entry they appear in.
Interaction-as-a-Judge
According to the description, a preprint proposes having computer-use agents interact with a generated application to judge it, instead of relying on code analysis and unit tests. The authors release two datasets and report better prediction of human preferences (65.6% to 69.1%) and requirement verification (77.0% to 81.6%) than non-interactive approaches.