Code and Data: Design and Realization of High Performance Textured Lead-Free Piezoelectric Ceramics through Human-AI Collaboration

Le résumé fourni par la source

Code and data for the paper "Design and Realization of High Performance Textured Lead-Free Piezoelectric Ceramics through Human-AI Collaboration". This repository contains: LLM-based composition generation scripts using Claude 3.5 Sonnet for designing new KNN-based lead-free piezoelectric ceramics Post-hoc statistical analysis code (Leave-One-Out, Conditional Prediction, Incremental Learning) using DeepSeek-V3 via Together AI Prediction benchmarks across multiple LLMs (Claude, GPT, Gemini, DeepSeek) All pre-generated data and visualization scripts for reproducing figures in the paper and supplementary information A Google Colab notebook for one-click reproduction of all plots (no API keys required)

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.