LLM-agent framework DyRIS improves space group prediction for imbalanced perovskite data

Researchers have developed DyRIS, a novel LLM-agent framework designed to predict the space groups of double perovskites, a class of materials with significant catalytic potential. Existing datasets for this task are often imbalanced, favoring common space groups over rarer ones. DyRIS addresses this by employing dynamic, diversity-enhanced few-shot prompting to retrieve relevant examples and incorporating rule-guided inference based on crystallographic domain knowledge to refine predictions. This approach improves accuracy, particularly for underrepresented space groups, outperforming existing composition-based and descriptor-based methods. AI

IMPACT Enhances LLM capabilities in scientific research, particularly for imbalanced datasets in materials science.

RANK_REASON Academic paper detailing a new method for materials science using LLMs. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• CrabNet
• DyRIS
• LLM
• PyCaret

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →