Search papers, labs, and topics across Lattice.
2
0
5
0
LLM judges may recognize important reasoning steps, but they still miss critical nuances, highlighting a gap between legibility and true interpretability in AI reasoning.
Larger LLMs may ignore user context more than smaller models, revealing critical insights into how linguistic framing can sway model responses.
