Search papers, labs, and topics across Lattice.
Affiliation:
2
0
5
Evaluated on recommendation models, Auto-RecSys significantly reduces the human time required per experiment cycle and improves execution reliability as its playbooks mature.
Long-horizon LLM agents can achieve 96.9% task success by learning to adapt their external execution support through trainable harness policies.
