vector-institute/s2ef-15m

该数据集包含从多个来源收集的3D原子数据集，带有力和能量标签。数据集的结构包括每个实例的原子编号（input_ids）、3D坐标（coords）、每个原子的力（forces）、系统的总能量和形成能量（total_energy/formation_energy）以及一个布尔值（has_formation_energy）表示数据集是否具有有效的形成能量。数据集还通过上采样确保了数据分布的平衡。

This dataset contains a collection of 3D atomistic datasets with force and energy labels gathered from a series of sources. The dataset structure includes atomic numbers (input_ids), 3-D coordinates (coords), forces per atom (forces), total and formation energy per system (total_energy/formation_energy), and a boolean (has_formation_energy) indicating whether the dataset has a valid formation energy. The datasets are balanced through upsampling to ensure a balanced distribution.

原始信息汇总

数据集概述

数据集描述

该数据集包含一系列3D原子数据集，带有力与能量标签，来源于以下几个来源：

• Open Catalyst Project (OC20, OC22, ODAC23)
• Materials Project Trajectory Dataset (MPtrj)
• SPICE 1.1.4

数据结构

数据实例

每个实例包含以下特征：

• input_ids: 原子编号序列，类型为int16。
• coords: 3D坐标序列，类型为float32。
• forces: 每个原子的力序列，类型为float32。
• formation_energy: 形成能，类型为float32。
• total_energy: 总能量，类型为float32。
• has_formation_energy: 是否具有有效的形成能，类型为bool。

示例数据： json { "input_ids": [26, 28, 28, 28], "coords": [[0.0, 0.0, 0.0], [0.0, 0.0, 3.5395920276641846], [0.0, 1.7669789791107178, 1.7697960138320923], [1.7669789791107178, 0.0, 1.7697960138320923]], "forces": [[-1.999999987845058e-08, 2.999999892949745e-08, -0.0], [-5.99999978589949e-08, 5.99999978589949e-08, 9.99999993922529e-09], [-0.0014535699738189578, 0.0014535400550812483, 9.99999993922529e-09], [0.001453649951145053, -0.0014536300441250205, -2.999999892949745e-08]], "formation_energy": 0.6030612587928772, "total_energy": -25.20570182800293, "has_formation_energy": true }

数据集划分

• train: 包含15,000,000个样本，数据大小为43,353,603,080字节。

数据集大小

• 下载大小: 44,763,791,790字节
• 数据集大小: 43,353,603,080字节

引用信息

plaintext @article{ocp_dataset, author = {Chanussot*, Lowik and Das*, Abhishek and Goyal*, Siddharth and Lavril*, Thibaut and Shuaibi*, Muhammed and Riviere, Morgane and Tran, Kevin and Heras-Domingo, Javier and Ho, Caleb and Hu, Weihua and Palizhati, Aini and Sriram, Anuroop and Wood, Brandon and Yoon, Junwoong and Parikh, Devi and Zitnick, C. Lawrence and Ulissi, Zachary}, title = {Open Catalyst 2020 (OC20) Dataset and Community Challenges}, journal = {ACS Catalysis}, year = {2021}, doi = {10.1021/acscatal.0c04525}, }

@article{oc22_dataset, author = {Tran*, Richard and Lan*, Janice and Shuaibi*, Muhammed and Wood*, Brandon and Goyal*, Siddharth and Das, Abhishek and Heras-Domingo, Javier and Kolluru, Adeesh and Rizvi, Ammar and Shoghi, Nima and Sriram, Anuroop and Ulissi, Zachary and Zitnick, C. Lawrence}, title = {The Open Catalyst 2022 (OC22) dataset and challenges for oxide electrocatalysts}, journal = {ACS Catalysis}, year={2023}, }

@article{odac23_dataset, author = {Anuroop Sriram and Sihoon Choi and Xiaohan Yu and Logan M. Brabson and Abhishek Das and Zachary Ulissi and Matt Uyttendaele and Andrew J. Medford and David S. Sholl}, title = {The Open DAC 2023 Dataset and Challenges for Sorbent Discovery in Direct Air Capture}, year = {2023}, journal={arXiv preprint arXiv:2311.00341}, }

@article{deng_2023_chgnet, author={Deng, Bowen and Zhong, Peichen and Jun, KyuJung and Riebesell, Janosh and Han, Kevin and Bartel, Christopher J. and Ceder, Gerbrand}, title={CHGNet as a pretrained universal neural network potential for charge-informed atomistic modelling}, journal={Nature Machine Intelligence}, year={2023}, DOI={10.1038/s42256-023-00716-3}, pages={1–11} }

@article{eastman2023spice, title={Spice, a dataset of drug-like molecules and peptides for training machine learning potentials}, author={Eastman, Peter and Behara, Pavan Kumar and Dotson, David L and Galvelis, Raimondas and Herr, John E and Horton, Josh T and Mao, Yuezhi and Chodera, John D and Pritchard, Benjamin P and Wang, Yuanqing and others}, journal={Scientific Data}, volume={10}, number={1}, pages={11}, year={2023}, publisher={Nature Publishing Group UK London} }