1 paper
Jiexing Qi, Yu He, Jun Liu +9
Language model agents can be improved by updating their model weights or refining the harness that guides task execution. These components are coupled: weight updates change how th…