Randall Balestriero

9

Papers

18

Total Citations

Papers (9)

$\mathbb{X}$-Sample Contrastive Loss: Improving Contrastive Learning with Sample Similarity Graphs

Beyond [cls]: Exploring the True Potential of Masked Image Modeling Representations

Ditch the Denoiser: Emergence of Noise Robustness in Self-Supervised Learning from Data Curriculum

Curvature Tuning: Provable Training-free Model Steering From a Single Parameter

From Linearity to Non-Linearity: How Masked Autoencoders Capture Spatial Correlations

Characterizing Large Language Model Geometry Helps Solve Toxicity Detection and Generation

FastDINOv2: Frequency Based Curriculum Learning Improves Robustness and Training Speed

How Learning by Reconstruction Produces Uninformative Features For Perception

Deep Networks Always Grok and Here is Why