Behemoth: Benchmarking Unlearning in LLMs Using Fully Synthetic Data

• URL: http://arxiv.org/abs/2601.23153v1
• Date: Fri, 30 Jan 2026 16:39:42 GMT
• Title: Behemoth: Benchmarking Unlearning in LLMs Using Fully Synthetic Data
• Authors: Eugenia Iofinova, Dan Alistarh,
• Abstract summary: We propose Behemoth, a framework for understanding the effects of model editing on large language models trained on real-world data.<n>We show that, in some cases, echo real-world results, for instance, that in some cases restricting the update rank results in a more effective update.
• Score: 43.026389128544594
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: As artificial neural networks, and specifically large language models, have improved rapidly in capabilities and quality, they have increasingly been deployed in real-world applications, from customer service to Google search, despite the fact that they frequently make factually incorrect or undesirable statements. This trend has inspired practical and academic interest in model editing, that is, in adjusting the weights of the model to modify its likely outputs for queries relating to a specific fact or set of facts. This may be done either to amend a fact or set of facts, for instance, to fix a frequent error in the training data, or to suppress a fact or set of facts entirely, for instance, in case of dangerous knowledge. Multiple methods have been proposed to do such edits. However, at the same time, it has been shown that such model editing can be brittle and incomplete. Moreover the effectiveness of any model editing method necessarily depends on the data on which the model is trained, and, therefore, a good understanding of the interaction of the training data distribution and the way it is stored in the network is necessary and helpful to reliably perform model editing. However, working with large language models trained on real-world data does not allow us to understand this relationship or fully measure the effects of model editing. We therefore propose Behemoth, a fully synthetic data generation framework. To demonstrate the practical insights from the framework, we explore model editing in the context of simple tabular data, demonstrating surprising findings that, in some cases, echo real-world results, for instance, that in some cases restricting the update rank results in a more effective update. The code is available at https://github.com/IST-DASLab/behemoth.git.

Related papers

• The Mirage of Model Editing: Revisiting Evaluation in the Wild [70.17413507444704]
  We introduce QAEdit, a new benchmark aligned with widely used question answering (QA) datasets, and WILD, a task-agnostic evaluation framework.<n>Our single editing experiments show that current editing methods perform substantially worse than previously reported.
  arXiv  Detail & Related papers  (2025-02-16T15:57:55Z)
• FAME: Towards Factual Multi-Task Model Editing [4.858226284963096]
  Large language models (LLMs) embed extensive knowledge and utilize it to perform exceptionally well across various tasks. We present FAME, an factual, comprehensive, and multi-task dataset, which is designed to enhance the practicality of model editing. We then propose SKEME, a model editing method that uses a novel caching mechanism to ensure synchronization with the real world.
  arXiv  Detail & Related papers  (2024-10-07T13:46:06Z)
• Better Call SAUL: Fluent and Consistent Language Model Editing with Generation Regularization [48.07144492109635]
  Large language models need to be updated regularly. Model editing is challenging as it might also affect knowledge that is unrelated to the new data. We propose SAUL, a streamlined model editing method that uses sentence concatenation with augmented random facts for generation regularization.
  arXiv  Detail & Related papers  (2024-10-03T12:28:13Z)
• The Butterfly Effect of Model Editing: Few Edits Can Trigger Large Language Models Collapse [58.0132400208411]
  Even a single edit can trigger model collapse, manifesting as significant performance degradation in various benchmark tasks. benchmarking Large Language Models after each edit is impractically time-consuming and resource-intensive. We have utilized GPT-3.5 to develop a new dataset, HardEdit, based on hard cases.
  arXiv  Detail & Related papers  (2024-02-15T01:50:38Z)
• Does Localization Inform Editing? Surprising Differences in Causality-Based Localization vs. Knowledge Editing in Language Models [68.03946716358335]
  We find that we can change how a fact is stored in a model by editing weights that are in a different location than where existing methods suggest that the fact is stored. This is surprising because we would expect that localizing facts to specific model parameters would tell us where to manipulate knowledge in models. Our results suggest, counterintuitively, that better mechanistic understanding of how pretrained language models work may not always translate to insights about how to best change their behavior.
  arXiv  Detail & Related papers  (2023-01-10T21:26:08Z)
• Memory-Based Model Editing at Scale [102.28475739907498]
  Existing model editors struggle to accurately model an edit's intended scope. We propose Semi-Parametric Editing with a Retrieval-Augmented Counterfactual Model (SERAC) SERAC stores edits in an explicit memory and learns to reason over them to modulate the base model's predictions as needed.
  arXiv  Detail & Related papers  (2022-06-13T23:40:34Z)
• Fast Model Editing at Scale [77.69220974621425]
  We propose Model Editor Networks with Gradient Decomposition (MEND) MEND is a collection of small auxiliary editing networks that use a single desired input-output pair to make fast, local edits to a pre-trained model. MEND can be trained on a single GPU in less than a day even for 10 billion+ parameter models.
  arXiv  Detail & Related papers  (2021-10-21T17:41:56Z)

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.