Search papers, labs, and topics across Lattice.
3
0
7
2
The implementation lottery reveals that relying on a single run can mislead research conclusions, with winner reversals occurring in up to 43.6% of cases.
Continual learning for LLM agents hits a wall: scaling models doesn't reliably improve skill generation, and self-feedback can lead to recursive drift.
Multi-LLM revision pipelines often succeed not because of error correction, but because the second model simply re-solves the problem, especially in constrained tasks like multiple-choice question answering.
