Advancing Adversarial Suffix Transfer Learning on Aligned Large Language Models

• URL: http://arxiv.org/abs/2408.14866v2
• Date: Sat, 5 Oct 2024 07:52:15 GMT
• Title: Advancing Adversarial Suffix Transfer Learning on Aligned Large Language Models
• Authors: Hongfu Liu, Yuxi Xie, Ye Wang, Michael Shieh,
• Abstract summary: Language Language Models (LLMs) face safety concerns due to potential misuse by malicious users. Recent red-teaming efforts have identified adversarial suffixes capable of jailbreaking LLMs using the gradient-based search algorithm Greedy Coordinate Gradient (GCG) We propose a two-stage transfer learning framework, DeGCG, which decouples the search process into behavior-agnostic pre-searching and behavior-relevant post-searching.
• Score: 21.96773736059112
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Language Language Models (LLMs) face safety concerns due to potential misuse by malicious users. Recent red-teaming efforts have identified adversarial suffixes capable of jailbreaking LLMs using the gradient-based search algorithm Greedy Coordinate Gradient (GCG). However, GCG struggles with computational inefficiency, limiting further investigations regarding suffix transferability and scalability across models and data. In this work, we bridge the connection between search efficiency and suffix transferability. We propose a two-stage transfer learning framework, DeGCG, which decouples the search process into behavior-agnostic pre-searching and behavior-relevant post-searching. Specifically, we employ direct first target token optimization in pre-searching to facilitate the search process. We apply our approach to cross-model, cross-data, and self-transfer scenarios. Furthermore, we introduce an interleaved variant of our approach, i-DeGCG, which iteratively leverages self-transferability to accelerate the search process. Experiments on HarmBench demonstrate the efficiency of our approach across various models and domains. Notably, our i-DeGCG outperforms the baseline on Llama2-chat-7b with ASRs of $43.9$ ($+22.2$) and $39.0$ ($+19.5$) on valid and test sets, respectively. Further analysis on cross-model transfer indicates the pivotal role of first target token optimization in leveraging suffix transferability for efficient searching.

Related papers

• Sequencing Matters: A Generate-Retrieve-Generate Model for Building Conversational Agents [9.191944519634111]
  The Georgetown InfoSense group has done in regard to solving the challenges presented by TREC iKAT 2023. Our submitted runs outperform the median runs by a significant margin, exhibiting superior performance in nDCG across various cut numbers and in overall success rate. Our solution involves the use of Large Language Models (LLMs) for initial answers, answer grounding by BM25, passage quality filtering by logistic regression, and answer generation by LLMs again.
  arXiv  Detail & Related papers  (2023-11-16T02:37:58Z)
• Single-Stage Visual Relationship Learning using Conditional Queries [60.90880759475021]
  TraCQ is a new formulation for scene graph generation that avoids the multi-task learning problem and the entity pair distribution. We employ a DETR-based encoder-decoder conditional queries to significantly reduce the entity label space as well. Experimental results show that TraCQ not only outperforms existing single-stage scene graph generation methods, it also beats many state-of-the-art two-stage methods on the Visual Genome dataset.
  arXiv  Detail & Related papers  (2023-06-09T06:02:01Z)
• Self-Evaluation Guided Beam Search for Reasoning [61.523627290397556]
  We introduce a stepwise self-evaluation mechanism to guide and calibrate the reasoning process of Large Language Model (LLM) We propose a decoding algorithm integrating the self-evaluation guidance via beam search. Our approach surpasses the corresponding Codex-backboned baselines in few-shot accuracy by $6.34%$, $9.56%$, and $5.46%$ on the GSM8K, AQuA, and StrategyQA.
  arXiv  Detail & Related papers  (2023-05-01T02:37:59Z)
• Logit Margin Matters: Improving Transferable Targeted Adversarial Attack by Logit Calibration [85.71545080119026]
  Cross-Entropy (CE) loss function is insufficient to learn transferable targeted adversarial examples. We propose two simple and effective logit calibration methods, which are achieved by downscaling the logits with a temperature factor and an adaptive margin. Experiments conducted on the ImageNet dataset validate the effectiveness of the proposed methods.
  arXiv  Detail & Related papers  (2023-03-07T06:42:52Z)
• GFlowCausal: Generative Flow Networks for Causal Discovery [27.51595081346858]
  We propose a novel approach to learning a Directed Acyclic Graph (DAG) from observational data called GFlowCausal. GFlowCausal aims to learn the best policy to generate high-reward DAGs by sequential actions with probabilities proportional to predefined rewards. We conduct extensive experiments on both synthetic and real datasets, and results show the proposed approach to be superior and also performs well in a large-scale setting.
  arXiv  Detail & Related papers  (2022-10-15T04:07:39Z)
• Adaptive Fine-Grained Predicates Learning for Scene Graph Generation [122.4588401267544]
  General Scene Graph Generation (SGG) models tend to predict head predicates and re-balancing strategies prefer tail categories. We propose an Adaptive Fine-Grained Predicates Learning (FGPL-A) which aims at differentiating hard-to-distinguish predicates for SGG. Our proposed model-agnostic strategy significantly boosts performance of benchmark models on VG-SGG and GQA-SGG datasets by up to 175% and 76% on Mean Recall@100, achieving new state-of-the-art performance.
  arXiv  Detail & Related papers  (2022-07-11T03:37:57Z)
• Training Generative Adversarial Networks with Adaptive Composite Gradient [2.471982349512685]
  This paper proposes the adaptive Composite Gradients (ACG) method, linearly convergent in bilinear games. ACG is a semi-gradient-free algorithm since it does not need to calculate the gradient of each step. Results show ACG is competitive with the previous algorithms.
  arXiv  Detail & Related papers  (2021-11-10T03:13:53Z)
• TraND: Transferable Neighborhood Discovery for Unsupervised Cross-domain Gait Recognition [77.77786072373942]
  This paper proposes a Transferable Neighborhood Discovery (TraND) framework to bridge the domain gap for unsupervised cross-domain gait recognition. We design an end-to-end trainable approach to automatically discover the confident neighborhoods of unlabeled samples in the latent space. Our method achieves state-of-the-art results on two public datasets, i.e., CASIA-B and OU-LP.
  arXiv  Detail & Related papers  (2021-02-09T03:07:07Z)
• Diverse Knowledge Distillation for End-to-End Person Search [81.4926655119318]
  Person search aims to localize and identify a specific person from a gallery of images. Recent methods can be categorized into two groups, i.e., two-step and end-to-end approaches. We propose a simple yet strong end-to-end network with diverse knowledge distillation to break the bottleneck.
  arXiv  Detail & Related papers  (2020-12-21T09:04:27Z)

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.