Can a Model Catch Its Own Hallucinations for Free?: Label-Free Doubt Signals Hold Their Own Against a Labelled Dataset for Abstention
Source
arxiv.org
Author
Ali Asaria, Tony Salomone, Deep Gandhi
Date
Why it matters
Abstention tuning may not require a labelled right/wrong dataset — a model's own tokenThe chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.Full definition → confidence is a free supervision signal that held its own against labelled baselines at matched coverage on short-form factual QA.
Terms in this piece · Glossary
fine-tuning — Taking a trained model and training it a bit more on your own examples so it gets better at one specific job.
open weights — A model whose trained parameters are published for anyone to download and run — unlike API-only models you can access but never possess.
token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
articleRefusing the Impossible: A Taxonomy and Benchmark for Code Hallucination in Large Language ModelsVishnu Asutosh Dasu, Ashish Kundu, Gang Tan
articleWhen Is a Multi-Agent Code Judge Actually Grounded? Two Label-Free Measurements, and a Judge That Declines to GuessSalma Roshdy Aly, Hussein Assaf, Ziad Kobti
articleBeyond the Sycophancy Score: How Task, Model, and Pressure Shape LLM YieldingGuang Yang, Homa Hosseinmardi, Fengchen Liu, Amir Ghasemian