Saved in:
Bibliographic Details
Main Authors: Gilligan, Luke P. J., Cobelli, Matteo, Sayeed, Hasan M., Sparks, Taylor D., Sanvito, Stefano
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2409.11971
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916399799074816
author Gilligan, Luke P. J.
Cobelli, Matteo
Sayeed, Hasan M.
Sparks, Taylor D.
Sanvito, Stefano
author_facet Gilligan, Luke P. J.
Cobelli, Matteo
Sayeed, Hasan M.
Sparks, Taylor D.
Sanvito, Stefano
contents Vector embeddings derived from large language models (LLMs) show promise in capturing latent information from the literature. Interestingly, these can be integrated into material embeddings, potentially useful for data-driven predictions of materials properties. We investigate the extent to which LLM-derived vectors capture the desired information and their potential to provide insights into material properties without additional training. Our findings indicate that, although LLMs can be used to generate representations reflecting certain property information, extracting the embeddings requires identifying the optimal contextual clues and appropriate comparators. Despite this restriction, it appears that LLMs still have the potential to be useful in generating meaningful materials-science representations.
format Preprint
id arxiv_https___arxiv_org_abs_2409_11971
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Sampling Latent Material-Property Information From LLM-Derived Embedding Representations
Gilligan, Luke P. J.
Cobelli, Matteo
Sayeed, Hasan M.
Sparks, Taylor D.
Sanvito, Stefano
Computation and Language
Materials Science
Vector embeddings derived from large language models (LLMs) show promise in capturing latent information from the literature. Interestingly, these can be integrated into material embeddings, potentially useful for data-driven predictions of materials properties. We investigate the extent to which LLM-derived vectors capture the desired information and their potential to provide insights into material properties without additional training. Our findings indicate that, although LLMs can be used to generate representations reflecting certain property information, extracting the embeddings requires identifying the optimal contextual clues and appropriate comparators. Despite this restriction, it appears that LLMs still have the potential to be useful in generating meaningful materials-science representations.
title Sampling Latent Material-Property Information From LLM-Derived Embedding Representations
topic Computation and Language
Materials Science
url https://arxiv.org/abs/2409.11971