Search papers, labs, and topics across Lattice.
3
0
7
A transparent probe-success rule boosts robot policy selection success rates by over 14 percentage points, revealing the hidden power of pre-deployment evaluations.
Agon reveals that machine-driven research can scale effectively while exposing critical failure modes that still require human oversight.
Multi-agent orchestration prompting is critically under-evaluated, with only 14.9% of models passing the new PerspectiveGap benchmark.
