From Text to Insight: Large Language Models for Materials Science Data Extraction
第一作者:Schilling-Wilhelmi M, Ríos-García M
作者单位:Friedrich Schiller University Jena, Institute of Carbon Science and Technology (INCAR)
发表时间:2024/7
发表期刊:
关键内容:对材料科学中基于LLM的结构化数据抽取进行了全面的综述,综合了当前的知识并概述了未来的发展方向。
Dealing with finite context: 处理文本长度,因为上下文窗口长度有限。
Beyond text: 除文本外其它类型数据。
1. Overview of the working principles of LLMs
对于结构化数据提取任务,温度值为0的工作通常是最好的,因为这将导致具有最相关信息的确定性输出。
2. Structured data extraction workflow
“A simple example of a system prompt for the data extraction task can be: “You are a material expert assistant and your task is to extract certain information from a given text fragment. If no information is provided for some variables, return NULL”.”
3. 参考文献
Schilling-Wilhelmi M, Ríos-García M, Shabih S, et al. From Text to Insight: Large Language Models for Materials Science Data Extraction[J]. arXiv preprint arXiv:2407.16867, 2024.
All articles in this blog are licensed under CC BY-NC-SA 4.0 unless stating additionally.