ICSFSurvey
Explore concepts like Self-Correct, Self-Refine, Self-Improve, Self-Contradict, Self-Play, and Self-Knowledge, alongside o1-like reasoning elevation🍓 and hallu
What is ICSFSurvey?
Explore concepts like Self-Correct, Self-Refine, Self-Improve, Self-Contradict, Self-Play, and Self-Knowledge, alongside o1-like reasoning elevation🍓 and hallucination alleviation🍄.
Overview of ICSFSurvey
What it is
ICSFSurvey is the GitHub repository for the survey paper titled Internal Consistency and Self-Feedback in Large Language Models: A Survey. The README states that the survey aims to provide a unified perspective on self-evaluation and self-updating mechanisms in large language models, encapsulated within the frameworks of Internal Consistency and Self-Feedback. The repository covers concepts including Self-Correct, Self-Refine, Self-Improve, Self-Contradict, Self-Play, and Self-Knowledge, alongside o1-like reasoning elevation and hallucination alleviation.
Key features
- Includes expt-consistency-types, described as code and results for measuring consistency at different levels.
- Includes expt-gpt4o-responses, described as results from five different GPT-4o responses to the same query.
- Provides a paper list of references related to the survey, organized into sections such as Consistency Signal Acquisition, Reasoning Elevation, Hallucination Alleviation, Other Tasks, Meta Evaluation, and Theoretical Perspectives.
- Lists related survey papers and references across topics including confidence estimation, hallucination detection, uncertainty estimation, verbal critiquing, faithfulness measurement, and consistency estimation.
- README states that the paper list will be updated regularly.
- README includes citation information via CITATION.bib and news entries about paper versions and updates.
Use cases
- Finding references and related work on internal consistency and self-feedback in large language models.
- Studying consistency measurement code and results at different levels.
- Comparing multiple GPT-4o responses to the same query.
- Locating papers on reasoning elevation, hallucination alleviation, meta evaluation, and other listed tasks.
- Supporting citation of the survey through the provided citation entry.
Getting started
The README does not provide install or quick-start commands. See the repository README for available resources and instructions.
License and pricing
The metadata labels the pricing type as open_source. No license is stated in the metadata or README; the license is not stated. No pricing information is provided.
Based on the project's public README and repository metadata.