10 papers
Gauss-Newton Unlearning for the LLM Era
Lev McKinney, Anvith Thudi, Juhan Bae +4
Standard large language model training can create models that produce outputs their trainer deems unacceptable in deployment. The probability of these outputs can be reduced using…
IF-GUIDE: Influence Function-Guided Detoxification of LLMs
Zachary Coalson, Juhan Bae, Nicholas Carlini +1
We study how training data contributes to the emergence of toxic behaviors in large language models. Most prior work on reducing model toxicity adopts reactive approaches, such as…
Exploring Training Data Attribution under Limited Access Constraints
Shiyuan Zhang, Junwei Deng, Juhan Bae +1
Training data attribution (TDA) plays a critical role in understanding the influence of individual training data points on model predictions. Gradient-based TDA methods, popularize…
Better Training Data Attribution via Better Inverse Hessian-Vector Products
Andrew Wang, Elisa Nguyen, Runshi Yang +3
Training data attribution (TDA) provides insights into which training data is responsible for a learned model behavior. Gradient-based TDA methods such as influence functions and u…
Benchmarking Neural Network Training Algorithms
George E. Dahl, Frank Schneider, Zachary Nado +22
Training algorithms, broadly construed, are an essential part of every deep learning pipeline. Training algorithm improvements that speed up training across a wide variety of workl…
Influence Functions for Scalable Data Attribution in Diffusion Models
Bruno Mlodozeniec, Runa Eschenhagen, Juhan Bae +3
Diffusion models have led to significant advancements in generative modelling. Yet their widespread adoption poses challenges regarding data attribution and interpretability. In th…