1 paper · 1 filter
Yaorui Shi, Yuchun Miao, Yuxin Chen +5
The harness surrounding a language model is a central determinant of agent performance. Recent methods optimize harnesses by searching over complete programs, where individual mech…