Sparsity-Preserving Differentially Private Training of Large Embedding Models Badih Ghazi Google Research Mountain View, CA

Open in new window