Zobrazeno 1 - 10
of 44
pro vyhledávání: '"Andriushchenko, Maksym"'
Refusal training is widely used to prevent LLMs from generating harmful, undesirable, or illegal outputs. We reveal a curious generalization gap in the current refusal training approaches: simply reformulating a harmful request in the past tense (e.g
Externí odkaz:
http://arxiv.org/abs/2407.11969
Autor:
Zou, Andy, Phan, Long, Wang, Justin, Duenas, Derek, Lin, Maxwell, Andriushchenko, Maksym, Wang, Rowan, Kolter, Zico, Fredrikson, Matt, Hendrycks, Dan
AI systems can take harmful actions and are highly vulnerable to adversarial attacks. We present an approach, inspired by recent advances in representation engineering, that interrupts the models as they respond with harmful outputs with "circuit bre
Externí odkaz:
http://arxiv.org/abs/2406.04313
In-context learning (ICL) allows LLMs to learn from examples without changing their weights, which is a particularly promising capability for long-context LLMs that can potentially learn from many examples. Recently, Lin et al. (2024) proposed URIAL,
Externí odkaz:
http://arxiv.org/abs/2405.19874
Autor:
Rando, Javier, Croce, Francesco, Mitka, Kryštof, Shabalin, Stepan, Andriushchenko, Maksym, Flammarion, Nicolas, Tramèr, Florian
Large language models are aligned to be safe, preventing users from generating harmful content like misinformation or instructions for illegal activities. However, previous work has shown that the alignment process is vulnerable to poisoning attacks.
Externí odkaz:
http://arxiv.org/abs/2404.14461
We show that even the most recent safety-aligned LLMs are not robust to simple adaptive jailbreaking attacks. First, we demonstrate how to successfully leverage access to logprobs for jailbreaking: we initially design an adversarial prompt template (
Externí odkaz:
http://arxiv.org/abs/2404.02151
Autor:
Chao, Patrick, Debenedetti, Edoardo, Robey, Alexander, Andriushchenko, Maksym, Croce, Francesco, Sehwag, Vikash, Dobriban, Edgar, Flammarion, Nicolas, Pappas, George J., Tramer, Florian, Hassani, Hamed, Wong, Eric
Jailbreak attacks cause large language models (LLMs) to generate harmful, unethical, or otherwise objectionable content. Evaluating these attacks presents a number of challenges, which the current collection of benchmarks and evaluation techniques do
Externí odkaz:
http://arxiv.org/abs/2404.01318
There is a consensus that instruction fine-tuning of LLMs requires high-quality data, but what are they? LIMA (NeurIPS 2023) and AlpaGasus (ICLR 2024) are state-of-the-art methods for selecting such high-quality examples, either via manual curation o
Externí odkaz:
http://arxiv.org/abs/2402.04833
Autor:
Debenedetti, Edoardo, Wan, Zishen, Andriushchenko, Maksym, Sehwag, Vikash, Bhardwaj, Kshitij, Kailkhura, Bhavya
The last six years have witnessed significant progress in adversarially robust deep learning. As evidenced by the CIFAR-10 dataset category in RobustBench benchmark, the accuracy under $\ell_\infty$ adversarial perturbations improved from 44\% in \ci
Externí odkaz:
http://arxiv.org/abs/2312.13131
Training an overparameterized neural network can yield minimizers of different generalization capabilities despite the same level of training loss. Meanwhile, with evidence that suggests a strong correlation between the sharpness of minima and their
Externí odkaz:
http://arxiv.org/abs/2311.17539
Weight decay is a broadly used technique for training state-of-the-art deep networks, including large language models. Despite its widespread usage, its role remains poorly understood. In this work, we highlight that the role of weight decay in moder
Externí odkaz:
http://arxiv.org/abs/2310.04415