Techniques for creating DP synthetic TEXT data
Abstract
In this section we will cover 3 main methods for creating DP synthetic data: DP finetuning, DP Inference and Private Evolution. We will provide comparison of the methods an outline their ideal usecases. Additionally we will talk about ways one can evaluate the quality and fidelity of DP synthetic TEXT data prior to using the data in downstream tasks.
Video
Chat is not available.
Successful Page Load