A Data-Based Perspective on Transfer LearningWe present a framework for pinpointing the impact of the source datasets in transfer learning.
When does Bias Transfer in Transfer Learning?We demonstrate that biases from pre-trained models can persist even after fine-tuning.
Distilling Model Failures as Directions in Latent SpaceWe demonstrate how to distill patterns of model errors as directions in a latent space.
Uncovering Brittleness with DatamodelsIn the second part of our datamodels series, we use datamodels to identify and study a new form of model brittleness.
Missingness Bias in Model DebuggingWe demonstrate how current missingness approximations introduce biases into model debugging.
Predicting Predictions with DatamodelsIn the first part of our datamodels series, we introduce datamodeling and its (linear) instantiation on CIFAR-10.
Editing a ClassifierWe develop a methodology for directly editing the predictions rules of a pre-trained classifier with virtually no additional data collection.
Combining Diverse Feature PriorsWe explore how a diverse set of feature priors can be leveraged to improve model generalization.