Search papers, labs, and topics across Lattice.
Affiliation:
1
0
2
8
Task-specific harness evolution can boost frozen LLM performance by over 39%, revealing a new path for enhancing AI agents without modifying their core architecture.