Reinforcement Learning with Metacognitive Feedback Elicits Faithful Uncertainty Expression in LLMs

Reinforcement Learning with Metacognitive Feedback Elicits Faithful Uncertainty Expression in LLMs

Reinforcement learning with metacognitive feedback and metacognitive data selection improve large language model calibration by enabling accurate self-assessment of performance and…

Hugging Face · Daily Papers ·Gabrielle Kaili-May Liu, Avi Caciularu · ·▲ 16 upvotes

Este artigo está em destaque na seleção diária de papers do Hugging Face, curada pela comunidade de pesquisa em IA.

Autores: Gabrielle Kaili-May Liu, Avi Caciularu, Gal Yona, Idan Szpektor, Arman Cohan

  • 16 upvotes da comunidade
  • Temas: reinforcement learning, metacognitive feedback, metacognitive data selection, faithful calibration, self-reported confidence scores, intrinsic uncertainty

Resumo

Resumo original (em inglês), extraído do paper:

Reinforcement learning with metacognitive feedback and metacognitive data selection improve large language model calibration by enabling accurate self-assessment of performance and uncertainty.

Onde ler

compartilhar: