Salvato in:
Dettagli Bibliografici
Autore principale: Smirnov, Roman
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:https://arxiv.org/abs/2412.06846
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916515079520256
author Smirnov, Roman
author_facet Smirnov, Roman
contents The paper describes LLM unlearning without a retaining dataset, using the ORPO reinforcement learning method with inference enhanced by modified classifier-free guidance. Significant improvement in unlearning, without degradation of the model, is achieved through direct training on synthetic replacement data in CFG-aware training regime, with classifier-free guidance applied during the inference. This article is an extended version of the NeurIPS 2024 LLM-PC submission, which was awarded second prize.
format Preprint
id arxiv_https___arxiv_org_abs_2412_06846
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Classifier-free guidance in LLMs Safety
Smirnov, Roman
Machine Learning
Artificial Intelligence
The paper describes LLM unlearning without a retaining dataset, using the ORPO reinforcement learning method with inference enhanced by modified classifier-free guidance. Significant improvement in unlearning, without degradation of the model, is achieved through direct training on synthetic replacement data in CFG-aware training regime, with classifier-free guidance applied during the inference. This article is an extended version of the NeurIPS 2024 LLM-PC submission, which was awarded second prize.
title Classifier-free guidance in LLMs Safety
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2412.06846