Saved in:
Bibliographic Details
Main Authors: Tikhonov, Alexey, Bylinina, Lisa, Yamshchikov, Ivan P.
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2409.18868
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912048529539072
author Tikhonov, Alexey
Bylinina, Lisa
Yamshchikov, Ivan P.
author_facet Tikhonov, Alexey
Bylinina, Lisa
Yamshchikov, Ivan P.
contents We show differences between a language-and-vision model CLIP and two text-only models - FastText and SBERT - when it comes to the encoding of individuation information. We study latent representations that CLIP provides for substrates, granular aggregates, and various numbers of objects. We demonstrate that CLIP embeddings capture quantitative differences in individuation better than models trained on text-only data. Moreover, the individuation hierarchy we deduce from the CLIP embeddings agrees with the hierarchies proposed in linguistics and cognitive science.
format Preprint
id arxiv_https___arxiv_org_abs_2409_18868
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Individuation in Neural Models with and without Visual Grounding
Tikhonov, Alexey
Bylinina, Lisa
Yamshchikov, Ivan P.
Computation and Language
Artificial Intelligence
Machine Learning
I.2.4; J.4; I.6.8; I.2.10
We show differences between a language-and-vision model CLIP and two text-only models - FastText and SBERT - when it comes to the encoding of individuation information. We study latent representations that CLIP provides for substrates, granular aggregates, and various numbers of objects. We demonstrate that CLIP embeddings capture quantitative differences in individuation better than models trained on text-only data. Moreover, the individuation hierarchy we deduce from the CLIP embeddings agrees with the hierarchies proposed in linguistics and cognitive science.
title Individuation in Neural Models with and without Visual Grounding
topic Computation and Language
Artificial Intelligence
Machine Learning
I.2.4; J.4; I.6.8; I.2.10
url https://arxiv.org/abs/2409.18868