arXiv cs.AI· Xinyue Zhang, Jixiang Li, Bin Shao, Baishun Yang, Zhiyang Liu, Weichao Wang·· 12 小时前AI 评分22
小语言模型微调实现可靠 VASP INCAR 生成，Qwen3-4B 超越 GPT-5.4
Fine-Tuning Small Language Models for Reliable VASP INCAR Generation
AI 导读
基于 Qwen3-4B 微调的小语言模型 INCAR-SLM 在 VASP INCAR 生成基准 INCARBench 上超越所有受测通用 LLM，百分制 INCAR Score 比 GPT-5.4 高 15.55 分。
正文 · 原文
Abstract:Language models can prepare VASP INCAR files from natural-language requests, but so far only large proprietary cloud models come close to handling the tightly coupled, physics-sensitive settings reliably, a dependence that fits poorly with local, high-throughput materials workflows where privacy, cost, and offline deployment matter. We show that a small language model (SLM) can close this gap. The SLM is fine-tuned on reference VASP calculations and paired with VASPGuard, a deterministic post-processor that checks syntax, workflow, and material-dependent constraints; we call the combined model INCAR-SLM. On INCARBench, a benchmark for VASP INCAR generation, INCAR-SLM built on Qwen3-4B outperforms every general-purpose LLM evaluated, exceeding GPT-5.4 by 15.55 points on the 100-point INCAR Score. Most of this gain comes from fine-tuning, with VASPGuard correcting the errors that remain. We further find that model size matters less than expected: once fine-tuning and post-processing are applied, performance saturates at a few billion parameters, and Qwen3-4B outperforms larger models in the same family.
| Subjects: | Materials Science (cond-mat.mtrl-sci); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2608.05387 [cond-mat.mtrl-sci] |
| (or arXiv:2608.05387v1 [cond-mat.mtrl-sci] for this version) | |
| https://doi.org/10.48550/arXiv.2608.05387 arXiv-issued DOI via DataCite |
Submission history
From: Bin Shao [view email]
[v1]
Wed, 5 Aug 2026 20:16:26 UTC (831 KB)
来源：arXiv cs.AI · arxiv.org