Search papers, labs, and topics across Lattice.
1
0
2
CHAIR reveals that analyzing internal logits can significantly boost hallucination detection accuracy in LLMs, especially in zero-shot contexts.
