NEP models for PTAF including PES and dipole

(This version is identical to the previous version 10.5281/zenodo.10255268 with the only change that the nep.txt files are added separately to provide easier interfacing with parsing scripts.) NEP models used in https://doi.org/10.48550/arXiv.2311.09739 and https://doi.org/10.48550/arXiv.2312.05233. The DFT data is obtained using the ASE ORCA interface with the following parameters: calc = ORCA(label='orcacalc', maxiter=500, charge=-1, mult=1, orcasimpleinput='PBE def2-TZVP TIGHTSCF engrad') train.xyz collects the relevant structures with associated information to train the NEP models on. pes.zip and dipole.zip include inputs (nep.in), final NEP models (nep.txt), as well as loss and scatter plots.(本版本与10.5281/zenodo.10255268的前一版本完全一致，仅新增了单独的nep.txt文件，以更便捷地对接解析脚本。) 本数据集所使用的神经进化势（NEP）模型引自下述两篇学术论文：https://doi.org/10.48550/arXiv.2311.09739 与 https://doi.org/10.48550/arXiv.2312.05233。本数据集的密度泛函理论（DFT, Density Functional Theory）数据通过ASE ORCA接口获取，具体参数设置如下：`calc = ORCA(label='orcacalc', maxiter=500, charge=-1, mult=1, orcasimpleinput='PBE def2-TZVP TIGHTSCF engrad')`。train.xyz文件包含用于训练该NEP模型的相关结构及其关联信息。pes.zip与dipole.zip压缩包中包含输入文件（nep.in）、最终的NEP模型文件（nep.txt），以及损失函数曲线与散点图。