Author

Pietro Lió

1 paper indexed here

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

#machine learning Preprint Sep 2026

Towards Identifying the Dataset Biases Causing Phantom Transfer

This work embeds a dataset's completions with Sentence-BERT, subtract the embeddings of clean reference completions, and compares the result to an open vocabulary of candidate topics, observing that different teacher models appear to express the same bias through different vocabulary.

Jonas Jürß, Pietro Lió · 0 citations