no code implementations • 23 Apr 2024 • Avi Schwarzschild, Zhili Feng, Pratyush Maini, Zachary C. Lipton, J. Zico Kolter
We outline the limitations of existing notions of memorization and show how the ACR overcomes these challenges by (i) offering an adversarial view to measuring memorization, especially for monitoring unlearning and compliance; and (ii) allowing for the flexibility to measure memorization for arbitrary strings at a reasonably low compute.
1 code implementation • 16 Apr 2024 • Yiming Zhang, Avi Schwarzschild, Nicholas Carlini, Zico Kolter, Daphne Ippolito
Despite being trained specifically to follow user instructions, today's language models perform poorly when instructed to produce random outputs.
1 code implementation • 4 Apr 2024 • Sean McLeish, Avi Schwarzschild, Tom Goldstein
We evaluate ChatGPT's ability to solve algorithm problems from the CLRS benchmark suite that is designed for GNNs.
1 code implementation • 22 Jan 2024 • Abhimanyu Hans, Avi Schwarzschild, Valeriia Cherepanova, Hamid Kazemi, Aniruddha Saha, Micah Goldblum, Jonas Geiping, Tom Goldstein
Detecting text generated by modern large language models is thought to be hard, as both LLMs and humans can exhibit a wide range of complex behaviors.
no code implementations • 11 Jan 2024 • Pratyush Maini, Zhili Feng, Avi Schwarzschild, Zachary C. Lipton, J. Zico Kolter
Large language models trained on massive corpora of data from the web can memorize and reproduce sensitive or private data raising both legal and ethical concerns.
no code implementations • 25 Nov 2023 • Sahil Verma, Gantavya Bhatt, Avi Schwarzschild, Soumye Singhal, Arnav Mohanty Das, Chirag Shah, John P Dickerson, Jeff Bilmes
In this work, we demonstrate that the efficacy of CleanCLIP in mitigating backdoors is highly dependent on the particular objective used during model pre-training.
3 code implementations • 9 Oct 2023 • Neel Jain, Ping-Yeh Chiang, Yuxin Wen, John Kirchenbauer, Hong-Min Chu, Gowthami Somepalli, Brian R. Bartoldson, Bhavya Kailkhura, Avi Schwarzschild, Aniruddha Saha, Micah Goldblum, Jonas Geiping, Tom Goldstein
We show that language model finetuning can be improved, sometimes dramatically, with a simple augmentation.
1 code implementation • 1 Sep 2023 • Neel Jain, Avi Schwarzschild, Yuxin Wen, Gowthami Somepalli, John Kirchenbauer, Ping-Yeh Chiang, Micah Goldblum, Aniruddha Saha, Jonas Geiping, Tom Goldstein
We find that the weakness of existing discrete optimizers for text, combined with the relatively high costs of optimization, makes standard adaptive attacks more challenging for LLMs.
no code implementations • 24 Apr 2023 • Randall Balestriero, Mark Ibrahim, Vlad Sobal, Ari Morcos, Shashank Shekhar, Tom Goldstein, Florian Bordes, Adrien Bardes, Gregoire Mialon, Yuandong Tian, Avi Schwarzschild, Andrew Gordon Wilson, Jonas Geiping, Quentin Garrido, Pierre Fernandez, Amir Bar, Hamed Pirsiavash, Yann Lecun, Micah Goldblum
Self-supervised learning, dubbed the dark matter of intelligence, is a promising path to advance machine learning.
no code implementations • 23 Mar 2023 • Avi Schwarzschild, Max Cembalest, Karthik Rao, Keegan Hines, John Dickerson
We observe on three datasets that we can train a model with this loss term to improve explanation consensus on unseen data, and see improved consensus between explainers other than those used in the loss term.
1 code implementation • 28 Feb 2023 • Alex Stein, Avi Schwarzschild, Michael Curry, Tom Goldstein, John Dickerson
It has been shown that neural networks can be used to approximate optimal mechanisms while satisfying the constraints that an auction be strategyproof and individually rational.
1 code implementation • 14 Feb 2023 • Arpit Bansal, Hong-Min Chu, Avi Schwarzschild, Soumyadip Sengupta, Micah Goldblum, Jonas Geiping, Tom Goldstein
Typical diffusion models are trained to accept a particular form of conditioning, most commonly text, and cannot be conditioned on other modalities without retraining.
1 code implementation • 30 Jun 2022 • Roman Levin, Valeriia Cherepanova, Avi Schwarzschild, Arpit Bansal, C. Bayan Bruss, Tom Goldstein, Andrew Gordon Wilson, Micah Goldblum
In this work, we demonstrate that upstream data gives tabular neural networks a decisive advantage over widely used GBDT models.
1 code implementation • 11 Feb 2022 • Arpit Bansal, Avi Schwarzschild, Eitan Borgnia, Zeyad Emam, Furong Huang, Micah Goldblum, Tom Goldstein
Algorithmic extrapolation can be achieved through recurrent systems, which can be iterated many times to solve difficult reasoning problems.
no code implementations • 29 Sep 2021 • Arpit Bansal, Avi Schwarzschild, Eitan Borgnia, Zeyad Emam, Furong Huang, Micah Goldblum, Tom Goldstein
Classical machine learning systems perform best when they are trained and tested on the same distribution, and they lack a mechanism to increase model power after training is complete.
1 code implementation • 13 Aug 2021 • Avi Schwarzschild, Eitan Borgnia, Arjun Gupta, Arpit Bansal, Zeyad Emam, Furong Huang, Micah Goldblum, Tom Goldstein
We describe new datasets for studying generalization from easy to hard examples.
no code implementations • 17 Jun 2021 • Arpit Bansal, Micah Goldblum, Valeriia Cherepanova, Avi Schwarzschild, C. Bayan Bruss, Tom Goldstein
Class-imbalanced data, in which some classes contain far more samples than others, is ubiquitous in real-world applications.
1 code implementation • NeurIPS 2021 • Avi Schwarzschild, Eitan Borgnia, Arjun Gupta, Furong Huang, Uzi Vishkin, Micah Goldblum, Tom Goldstein
In this work, we show that recurrent networks trained to solve simple problems with few recurrent steps can indeed solve much more complex problems simply by performing additional recurrences during inference.
7 code implementations • 2 Jun 2021 • Gowthami Somepalli, Micah Goldblum, Avi Schwarzschild, C. Bayan Bruss, Tom Goldstein
We devise a hybrid deep learning approach to solving tabular data problems.
1 code implementation • ICLR 2022 • Avi Schwarzschild, Arjun Gupta, Amin Ghiasi, Micah Goldblum, Tom Goldstein
It is widely believed that deep neural networks contain layer specialization, wherein neural networks extract hierarchical features representing edges and patterns in shallow layers and complete objects in deeper layers.
no code implementations • 1 Jan 2021 • Avi Schwarzschild, Micah Goldblum, Arjun Gupta, John P Dickerson, Tom Goldstein
Data poisoning and backdoor attacks manipulate training data in order to cause models to fail during inference.
no code implementations • 18 Dec 2020 • Micah Goldblum, Dimitris Tsipras, Chulin Xie, Xinyun Chen, Avi Schwarzschild, Dawn Song, Aleksander Madry, Bo Li, Tom Goldstein
As machine learning systems grow in scale, so do their training data requirements, forcing practitioners to automate and outsource the curation of training data in order to achieve state-of-the-art performance.
2 code implementations • 22 Jun 2020 • Avi Schwarzschild, Micah Goldblum, Arjun Gupta, John P. Dickerson, Tom Goldstein
Data poisoning and backdoor attacks manipulate training data in order to cause models to fail during inference.
no code implementations • 20 Apr 2020 • Ahmed Abdelkader, Michael J. Curry, Liam Fowl, Tom Goldstein, Avi Schwarzschild, Manli Shu, Christoph Studer, Chen Zhu
We first demonstrate successful transfer attacks against a victim network using \textit{only} its feature extractor.
no code implementations • 21 Feb 2020 • Micah Goldblum, Avi Schwarzschild, Ankit B. Patel, Tom Goldstein
Algorithmic trading systems are often completely automated, and deep learning is increasingly receiving attention in this domain.
1 code implementation • ICLR 2020 • Micah Goldblum, Jonas Geiping, Avi Schwarzschild, Michael Moeller, Tom Goldstein
We empirically evaluate common assumptions about neural networks that are widely held by practitioners and theorists alike.