Revisiting Who's Harry Potter: Towards Targeted Unlearning from a Causal Intervention Perspective

• URL: http://arxiv.org/abs/2407.16997v2
• Date: Mon, 7 Oct 2024 03:41:57 GMT
• Title: Revisiting Who's Harry Potter: Towards Targeted Unlearning from a Causal Intervention Perspective
• Authors: Yujian Liu, Yang Zhang, Tommi Jaakkola, Shiyu Chang,
• Abstract summary: We introduce a new task of LLM targeted unlearning, where given an unlearning target and some unlearning documents, we aim to unlearn only the information about the target. We argue that a successful unlearning should satisfy criteria such as not outputting gibberish, not fabricating facts about the unlearning target, and not releasing factual information under jailbreak attacks. This framework justifies and extends WHP, deriving a simple unlearning algorithm that includes WHP as a special case.
• Score: 32.93858075964824
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: This paper investigates Who's Harry Potter (WHP), a pioneering yet insufficiently understood method for LLM unlearning. We explore it in two steps. First, we introduce a new task of LLM targeted unlearning, where given an unlearning target (e.g., a person) and some unlearning documents, we aim to unlearn only the information about the target, rather than everything in the unlearning documents. We further argue that a successful unlearning should satisfy criteria such as not outputting gibberish, not fabricating facts about the unlearning target, and not releasing factual information under jailbreak attacks. Second, we construct a causal intervention framework for targeted unlearning, where the knowledge of the unlearning target is modeled as a confounder between LLM input and output, and the unlearning process as a deconfounding process. This framework justifies and extends WHP, deriving a simple unlearning algorithm that includes WHP as a special case. Experiments on existing and new datasets show that our approach, without explicitly optimizing for the aforementioned criteria, achieves competitive performance in all of them. Our code is available at https://github.com/UCSB-NLP-Chang/causal_unlearn.git.

Related papers

• Does Unlearning Truly Unlearn? A Black Box Evaluation of LLM Unlearning Methods [1.9799527196428242]
  Large language model unlearning aims to remove harmful information that LLMs have learnt to prevent their use for malicious purposes. LMU and RMU have been proposed as two methods for LLM unlearning, achieving impressive results on unlearning benchmarks.
  arXiv  Detail & Related papers  (2024-11-18T22:31:17Z)
• Evaluating Deep Unlearning in Large Language Models [26.01778651411487]
  We investigate whether current unlearning methods for large language models succeed beyond superficial unlearning of facts. We design the metric, recall, to quantify the extent of deep unlearning. Our findings reveal that in the task of deep unlearning only a single fact, they either fail to properly unlearn with high recall, or end up unlearning many other irrelevant facts.
  arXiv  Detail & Related papers  (2024-10-19T16:40:08Z)
• A Closer Look at Machine Unlearning for Large Language Models [46.245404272612795]
  Large language models (LLMs) may memorize sensitive or copyrighted content, raising privacy and legal concerns. We discuss several issues in machine unlearning for LLMs and provide our insights on possible approaches.
  arXiv  Detail & Related papers  (2024-10-10T16:56:05Z)
• UnUnlearning: Unlearning is not sufficient for content regulation in advanced generative AI [50.61495097098296]
  We revisit the paradigm in which unlearning is used for Large Language Models (LLMs) We introduce a concept of ununlearning, where unlearned knowledge gets reintroduced in-context. We argue that content filtering for impermissible knowledge will be required and even exact unlearning schemes are not enough for effective content regulation.
  arXiv  Detail & Related papers  (2024-06-27T10:24:35Z)
• Jogging the Memory of Unlearned LLMs Through Targeted Relearning Attacks [37.061187080745654]
  We show that existing approaches for unlearning in LLMs are surprisingly susceptible to a simple set of targeted relearning attacks. With access to only a small and potentially loosely related set of data, we find that we can "jog" the memory of unlearned models to reverse the effects of unlearning.
  arXiv  Detail & Related papers  (2024-06-19T09:03:21Z)
• Reversing the Forget-Retain Objectives: An Efficient LLM Unlearning Framework from Logit Difference [39.29939437034823]
  We propose a novel unlearning framework called Unlearning from Logit Difference (ULD) Our method efficiently achieves the intended forgetting while preserving the LLM's overall capabilities, reducing training time by more than threefold.
  arXiv  Detail & Related papers  (2024-06-12T19:26:35Z)
• Rethinking Machine Unlearning for Large Language Models [85.92660644100582]
  We explore machine unlearning in the domain of large language models (LLMs) This initiative aims to eliminate undesirable data influence (e.g., sensitive or illegal information) and the associated model capabilities.
  arXiv  Detail & Related papers  (2024-02-13T20:51:58Z)
• TOFU: A Task of Fictitious Unlearning for LLMs [99.92305790945507]
  Large language models trained on massive corpora of data from the web can reproduce sensitive or private data raising both legal and ethical concerns. Unlearning, or tuning models to forget information present in their training data, provides us with a way to protect private data after training. We present TOFU, a benchmark aimed at helping deepen our understanding of unlearning.
  arXiv  Detail & Related papers  (2024-01-11T18:57:12Z)
• Knowledge-Aware Meta-learning for Low-Resource Text Classification [87.89624590579903]
  This paper studies a low-resource text classification problem and bridges the gap between meta-training and meta-testing tasks. We propose KGML to introduce additional representation for each sentence learned from the extracted sentence-specific knowledge graph.
  arXiv  Detail & Related papers  (2021-09-10T07:20:43Z)
• Fast Few-Shot Classification by Few-Iteration Meta-Learning [173.32497326674775]
  We introduce a fast optimization-based meta-learning method for few-shot classification. Our strategy enables important aspects of the base learner objective to be learned during meta-training. We perform a comprehensive experimental analysis, demonstrating the speed and effectiveness of our approach.
  arXiv  Detail & Related papers  (2020-10-01T15:59:31Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.