Download all files (1.36 MB)

Collection

Citation

Description:

The integration of artificial intelligence into various domains is rapidly increasing, with Large Language Models (LLMs) becoming more prevalent in numerous applications. This work is include in an overall project which aims to train an LLM specifically in the field of materials science. To assess the impact of this specialized training, it is essential to establish the baseline performance of existing LLMs in materials science. In this study, we evaluated 15 different LLMs using the MaScQA question answering (Q&A) benchmark. This benchmark comprises questions from the Graduate Aptitude Test in Engineering (GATE), tailored to test models' capabilities in answering questions related to the materials science and metallurgical engineering. Our results indicate that closed-source LLMs, such as Claude-3.5-Sonnet and GPT-4o, perform the best with an overall accuracy of 84%, when the open-source models, Llama3-70b and Phi3-14b, top at 56 and 43%, respectively. These findings provide a baseline for the raw capabilities of LLMs on Q&A tasks applied to materials science, and emphasise the substantial improvement that could be brought to open-source models via prompt engineering and fine-tuning strategies. We anticipate that this work could push the adoption of LLMs as valuable assistants in materials science, demonstrating their utilities in this specialised domain and related sub-domains.

Rights:

• Creative Commons BY Attribution 4.0 International
  

Keyword: Large language models, Materials science, Metallurgical Engineering

Date published: 2025-01-20

Publisher: Royal Society of Chemistry

Journal:

• Digital Discovery (ISSN: 2635098X)

Funding:

Manuscript type: Publisher's version (Version of record)

MDR DOI:

First published URL: https://doi.org/10.1039/d4dd00319e

Related item:

Other identifier(s):

Contact agent:

Updated at: 2025-01-21 12:30:37 +0900

Published on MDR: 2025-01-21 12:30:37 +0900