ICSFSurvey

ICSFSurvey

In depth study of internal consistency and self feedback in large language models

  • Provide a theoretical framework for internal consistency to explain reasoning gaps and hallucinations in LLMs.
  • Self feedback mechanisms, including self-evaluation and self-renewal, to enhance model response or the model itself.
  • Systematic classification research, categorizing tasks and work lines based on self feedback mechanisms.
  • Summarize evaluation methods and benchmark tests to assess the effectiveness of self feedback.
  • Explore key points such as whether self feedback is truly effective, and propose hypotheses such as the hourglass evolution of internal consistency.
  • Overview of future research directions on internal consistency and self feedback in LLMs.

Product Details

ICSFSurvey is a survey study on the internal consistency and self feedback of large language models. It provides a unified perspective on the self-assessment and self-renewal mechanisms of LLMs, including theoretical frameworks, systematic classifications, evaluation methods, future research directions, and more.