Yu Fan
LightNews LightNews
yu-fan-768.bsky.social
Yu Fan
@yu-fan-768.bsky.social
5 followers 20 following 0 posts
Posts Replies Media Videos
Reposted by Yu Fan
alexanderhoyle.bsky.social
Accepted to EMNLP (and more to come 👀)! The camera ready version is now online---very happy with how this turned out

arxiv.org/abs/2507.01234
alexanderhoyle.bsky.social Alexander Hoyle @alexanderhoyle.bsky.social · Jul 17
New preprint! Have you ever tried to cluster text embeddings from different sources, but the clusters just reproduce the sources? Or attempted to retrieve similar documents across multiple languages, and even multilingual embeddings return items in the same language?

Turns out there's an easy fix🧵
Barchart of number of items in four clusters of text embeddings, with colors showing the distribution of sources in each cluster. Caption: Clustering text embeddings from disparate sources (here, U.S. congressional bill summaries and senators’ tweets) can produce clusters where one source dominates (Panel A). Using linear erasure to remove the source information produces more evenly balanced clusters that maintain semantic coherence (Panel B; sampled items relate to immigration). Four random clusters of k-means shown (k=25), trained on a combined 5,000 samples from each dataset
September 24, 2025 at 3:21 PM
Reposted by Yu Fan
alexanderhoyle.bsky.social
New preprint! Have you ever tried to cluster text embeddings from different sources, but the clusters just reproduce the sources? Or attempted to retrieve similar documents across multiple languages, and even multilingual embeddings return items in the same language?

Turns out there's an easy fix🧵
Barchart of number of items in four clusters of text embeddings, with colors showing the distribution of sources in each cluster. Caption: Clustering text embeddings from disparate sources (here, U.S. congressional bill summaries and senators’ tweets) can produce clusters where one source dominates (Panel A). Using linear erasure to remove the source information produces more evenly balanced clusters that maintain semantic coherence (Panel B; sampled items relate to immigration). Four random clusters of k-means shown (k=25), trained on a combined 5,000 samples from each dataset
July 17, 2025 at 10:53 AM