Search papers, labs, and topics across Lattice.
3
0
3
7
Evolving rubrics from a single query can dramatically enhance LLM evaluation by eliminating reliance on external annotations and improving answer quality discrimination.
Reward models can be made more robust to spurious cues like length and sycophancy by explicitly training them to understand the *intent* behind a prompt.
Observational user feedback, often dismissed as too noisy and biased, can actually power effective RLHF with the right causal modeling, achieving a 49.2% gain on WildGuardMix.
