Fetching the paper…

Sparsity-Preserving Differentially Private Training of Large Embedding Models · Around