Unfamiliar Finetuning Examples Control How Language Models Hallucinate

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Kang, Katie, Wallace, Eric, Tomlin, Claire, Kumar, Aviral, Levine, Sergey
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866916264479293440
author Kang, Katie
Wallace, Eric
Tomlin, Claire
Kumar, Aviral
Levine, Sergey
author_facet Kang, Katie
Wallace, Eric
Tomlin, Claire
Kumar, Aviral
Levine, Sergey
contents Large language models are known to hallucinate when faced with unfamiliar queries, but the underlying mechanism that govern how models hallucinate are not yet fully understood. In this work, we find that unfamiliar examples in the models' finetuning data -- those that introduce concepts beyond the base model's scope of knowledge -- are crucial in shaping these errors. In particular, we find that an LLM's hallucinated predictions tend to mirror the responses associated with its unfamiliar finetuning examples. This suggests that by modifying how unfamiliar finetuning examples are supervised, we can influence a model's responses to unfamiliar queries (e.g., say ``I don't know''). We empirically validate this observation in a series of controlled experiments involving SFT, RL, and reward model finetuning on TriviaQA and MMLU. Our work further investigates RL finetuning strategies for improving the factuality of long-form model generations. We find that, while hallucinations from the reward model can significantly undermine the effectiveness of RL factuality finetuning, strategically controlling how reward models hallucinate can minimize these negative effects. Leveraging our previous observations on controlling hallucinations, we propose an approach for learning more reliable reward models, and show that they improve the efficacy of RL factuality finetuning in long-form biography and book/movie plot generation tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2403_05612
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Unfamiliar Finetuning Examples Control How Language Models Hallucinate
Kang, Katie
Wallace, Eric
Tomlin, Claire
Kumar, Aviral
Levine, Sergey
Machine Learning
Artificial Intelligence
Computation and Language
Large language models are known to hallucinate when faced with unfamiliar queries, but the underlying mechanism that govern how models hallucinate are not yet fully understood. In this work, we find that unfamiliar examples in the models' finetuning data -- those that introduce concepts beyond the base model's scope of knowledge -- are crucial in shaping these errors. In particular, we find that an LLM's hallucinated predictions tend to mirror the responses associated with its unfamiliar finetuning examples. This suggests that by modifying how unfamiliar finetuning examples are supervised, we can influence a model's responses to unfamiliar queries (e.g., say ``I don't know''). We empirically validate this observation in a series of controlled experiments involving SFT, RL, and reward model finetuning on TriviaQA and MMLU. Our work further investigates RL finetuning strategies for improving the factuality of long-form model generations. We find that, while hallucinations from the reward model can significantly undermine the effectiveness of RL factuality finetuning, strategically controlling how reward models hallucinate can minimize these negative effects. Leveraging our previous observations on controlling hallucinations, we propose an approach for learning more reliable reward models, and show that they improve the efficacy of RL factuality finetuning in long-form biography and book/movie plot generation tasks.
title Unfamiliar Finetuning Examples Control How Language Models Hallucinate
topic Machine Learning
Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2403.05612