# aLLoyM: a large language model for alloy phase diagram prediction

https://mdr.nims.go.jp/datasets/2bc9a70d-646f-41a2-952d-ccec287f5d6f

## File

- [s41524-026-01966-6 (2).pdf](https://mdr.nims.go.jp/filesets/d861db92-69d5-4099-b174-d5c56eb100e6/download) ([Detail](https://mdr.nims.go.jp/filesets/d861db92-69d5-4099-b174-d5c56eb100e6.md))

## Id

2bc9a70d-646f-41a2-952d-ccec287f5d6f

## Local identifier



## Visibility

open_to_public

## State

published

## Created at

2026-03-31T21:33:35.988544Z

## Updated at

2026-04-01T04:56:07.916952Z

## Published at

2026-04-01T07:26:13.142569Z

## Doi



## First published url

https://doi.org/10.1038/s41524-026-01966-6

## Date published

2026-01-22

## Recorded date published



## Resource type

journal_article

## Manuscript type

vor

## Collection



## Title

- title: 'aLLoyM: a large language model for alloy phase diagram prediction'
  title_type: original
  lang: en

## Description

- description: 'Large language models (LLMs) are general-purpose tools with wide-ranging
    applications, including in materials science. In this work, we introduce aLLoyM,
    a fine-tuned LLM specifically trained on alloy compositions, temperatures, and
    their corresponding phase information. To develop aLLoyM, we curated question-and-answer
    (Q&A) pairs for binary and ternary phase diagrams using the open-source Computational
    Phase Diagram Database (CPDDB) and assessments based on CALPHAD (CALculation of
    PHAse Diagrams). We fine-tuned Mistral, an open-source pre-trained LLM, for two
    distinct Q&A formats: multiple-choice and short-answer. Benchmark evaluations
    demonstrate that fine-tuning substantially enhances performance on multiple-choice
    phase diagram questions. Moreover, the short-answer model of aLLoyM can generate
    novel phase diagrams from its components alone, suggesting that it may aid the
    discovery of new materials systems. To promote further research and adoption,
    we have publicly released the short-answer fine-tuned version of aLLoyM, along
    with the complete benchmarking Q&A dataset, on Hugging Face.'
  description_type: abstract
  lang: und

## Creator

- name: Yuna Oikawa
  role: author
- name: Guillaume Deffrennes
  role: author
- name: Rintaro Shimayoshi
  role: author
- name: Taichi Abe
  role: author
  orcid: https://orcid.org/0000-0002-5065-0939
- name: Ryo Tamura
  role: author
  orcid: https://orcid.org/0000-0002-0349-358X
- name: Koji Tsuda
  role: author

## Contact agent



## Publisher

organization: Springer Science and Business Media LLC

## Managing organization



## Keyword

- subject: LLM
  schema: not_defined
- subject: phase diagram
  schema: not_defined

## Rights

- identifier: https://creativecommons.org/licenses/by-nc-nd/4.0/

## Other identifier(s)



## Data origin

- data_origin_type: other

## Embargo



## Journal

- title: npj Computational Materials
  issn: '20573960'
  volume: '12'
  article_number: '97'

## Conference



## Related item



## Funding

- identifier: 25KJ0870
  funder_name: Japan Society for the Promotion of Science
- identifier: 25K01492
  funder_name: Japan Society for the Promotion of Science
- funder_name: 'MEXT Program: Data Creation and Utilization Type Material Research
    and Development Project'
- identifier: JPMJCR21O2
  funder_name: Japan Science and Technology Agency,Japan

## Instrument



## Instrument operator



## Instrument managing organization



## Measurement method



## Specimen



## Chemical composition



## Structure for specimen



## Structural feature for specimen



## Specific property for specimen



## Process for specimen treatment



## Computational method



## Energy level/transition state



## Software



## Custom property



## Fileset

- id: d861db92-69d5-4099-b174-d5c56eb100e6
  filename: s41524-026-01966-6 (2).pdf
  content_type: application/pdf
  size: 3495737
  md5: 9c17763d089d7bea0c2a1e404b22b412

## Thumbnail

fileset_id: d861db92-69d5-4099-b174-d5c56eb100e6
filename: s41524-026-01966-6 (2).pdf