Search papers, labs, and topics across Lattice.
2
0
5
2
LLMs can autoformalize specs well enough to pass standard tests, but still fail on subtle edge cases 26% of the time, a risk missed by LLM-as-judge evaluations.
On-policy reward modeling with LLM judges not only unlocks significant performance gains on complex mathematical reasoning tasks, but also generalizes to improve performance on simpler numerical and multiple-choice benchmarks.
