Tom Sander

Titles as the sources gave them. In 1 entry, with 0 checked quotations.

In the Base

Every entry they appear in.

  1. ResearcharXivWrote it

    Watermark Basins: Black-Box Detection of Distilled LLMs

    According to the abstract, Chang, Elsahar, Sander and co-authors report a signal they call the watermark basin, which they say shows up in parameter space when a model is fine-tuned on watermarked teacher outputs. They introduce WM-Basin, a black-box detector of distillation that queries a suspect model through its API.