Skip to content
AI News HubLIVE
Original source2 min read

SteelAgent: LLM-Orchestrated, Physics-Informed Steel Property Prediction with Audited Generalization and Calibrated Uncertainty

Summary

SteelAgent couples metallurgical descriptors and machine-learning models with a language-model interface for steel property prediction, uncertainty reporting and inverse design. The study shows that the choice of held-out unit dramatically changes conclusions: gradient-boosting R² falls from 0.859 on a random split to 0.113 when steel families are held out, and coverage of nominal 90% prediction intervals drops from 91.0% to 37.6%. The effect of physics-derived features also depends on the evaluation. The authors argue for reporting the evaluation population alongside property estimates or compliance scores.

SourcearXiv Machine LearningAuthor: Aleksandr Volkov, Roman Sultimov, Mikhail Kuzin, Yury Maximov
SteelAgent: LLM-Orchestrated, Physics-Informed Steel Property Prediction with Audited Generalization and Calibrated Uncertainty
Report an error

The correction channel is not available yet. You can copy the article reference below for later.

Correction instructions
Read article

[Submitted on 3 Nov 2026]

Title:SteelAgent: LLM-Orchestrated, Physics-Informed Steel Property Prediction with Audited Generalization and Calibrated Uncertainty

View a PDF of the paper titled SteelAgent: LLM-Orchestrated, Physics-Informed Steel Property Prediction with Audited Generalization and Calibrated Uncertainty, by Aleksandr Volkov and 3 other authors

View PDF HTML (experimental)

Abstract:Steel property models can perform well on random test splits even when they extrapolate poorly to unfamiliar alloys. SteelAgent couples metallurgical descriptors and machine-learning models with a language-model interface for property prediction, uncertainty reporting and inverse design. We examine how the choice of held-out unit changes the conclusions drawn from SteelBench tensile-strength evaluations. For gradient boosting, pooled out-of-fold $R^2$ fell from 0.859 on a random split to 0.113 when steel families were held out. Empirical coverage of nominal 90% prediction intervals ranged from 91.0% under random splitting to 37.6% on held-out sources. The effect of physics-derived features depended on the evaluation. They increased gradient-boosting $R^2$ by 0.2534 in a 17-family holdout analysis and decreased it by 0.358 in a separate four-grade laboratory analysis. We report group-level errors and a synthetic regression experiment illustrating how calibration can deteriorate on held-out groups. These results motivate reporting the evaluation population alongside a property estimate or compliance score.

Comments: 10 pages, 3 tables. A shorter version of this work was published in the Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence (IJCAI-26), Demonstrations Track, pp. 8547-8550 (2026), doi:https://doi.org/10.24963/ijcai.2026/1000. Includes code for the synthetic experiment

Subjects:

Machine Learning (cs.LG); Materials Science (cond-mat.mtrl-sci)

Cite as: arXiv:2611.06953 [cs.LG]

(or arXiv:2611.06953v1 [cs.LG] for this version)

https://doi.org/10.48550/arXiv.2611.06953

arXiv-issued DOI via DataCite

Submission history

From: Yury Maximov [view email] [v1] Tue, 3 Nov 2026 20:08:01 UTC (13 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled SteelAgent: LLM-Orchestrated, Physics-Informed Steel Property Prediction with Audited Generalization and Calibrated Uncertainty, by Aleksandr Volkov and 3 other authors

View PDF

HTML (experimental)

TeX Source

view license

Ancillary-file links:

Ancillary files (details):

synthetic_coverage.py

Additional Features

Audio Summary

Current browse context:

cs.LG

new | recent | 2026-11

Change to browse by:

cond-mat cond-mat.mtrl-sci cs

References & Citations

SciX

Google Scholar

Semantic Scholar

Loading...

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)

Related Papers

Recommenders and Search Tools

Link to Influence Flower

Influence Flower (What are Influence Flowers?)

Core recommender toggle

CORE Recommender (What is CORE?)

IArxiv recommender toggle

IArxiv Recommender (What is IArxiv?)

Author

Venue

Institution

Topic

About arXivLabs

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)

Key points and analysis

Article intelligence

InvestorsAdvanced

Key points

  • SteelAgent integrates metallurgical descriptors, ML models, and an LLM interface for property prediction, uncertainty reporting, and inverse design.
  • Gradient-boosting R² fell from 0.859 on a random split to 0.113 when steel families were held out; nominal 90% prediction-interval coverage dropped from 91.0% to 37.6%.
  • Physics-derived features increased R² by 0.2534 in a 17-family holdout but decreased it by 0.358 in a four-grade laboratory analysis.
  • The authors advocate disclosing the evaluation population alongside performance estimates or compliance scores and provide code for a synthetic regression experiment.

Highlights and analysis are generated automatically and may contain errors. Check the original source.

More recent stories on this topic · Models