We provide the dataset including GPT-synthesized narratives based on the publicly accessible databases, Materials Project, JARVIS, OQMD, and AFLOW2. The dataset is designed to train future artificial intelligence (AI) models which aware the materials science.
Texts describe each material based on their properties to minimize the hallucination of chatbots. Furthermore, leveraging the chatbots' reasoning power, the texts also describe potential applications of each material. We believe that our dataset plays an important role in training future AI models.