Search papers, labs, and topics across Lattice.
3
0
5
50
Optimization gains only compound when regression control is integrated into the agent's learning process, revealing a critical factor for effective continual learning.
Early hidden states of LLMs can predict steering success with surprising accuracy, enabling efficient steering without exhaustive rollouts.
LLMs can be tricked into using specific tools over others simply by tweaking the tool's description, even if the tool is less suitable.