OMAT24
收藏资源简介：
Meta Open Materials 2024 (OMat24) 数据集是一个用于化学和材料科学研究的材料数据集合。该数据集包含非平衡结构的单点计算和结构弛豫，标记有总能量、力和应力。数据以ASE DB兼容的lmdb文件格式提供，并分为训练和验证集。每个集包含基于不同输入生成策略的多个子数据集。该数据集完全兼容Matbench Discovery基准测试集，并且不包含与WBM数据集重叠的任何结构。此外，README还提到了sAlex数据集，这是一个用于OMat模型微调的亚采样版本。README还提供了使用fairchem库读取数据的说明，并提供了支持渠道以解决任何问题。
Meta Open Materials 2024 (OMat24) dataset is a materials data collection for chemical and materials science research. This dataset contains single-point calculations and structural relaxations of non-equilibrium structures, labeled with total energy, forces, and stresses. The data is provided in an ASE DB-compatible LMDB file format, and is split into training and validation sets. Each of these sets contains multiple subdatasets based on different input generation strategies. This dataset is fully compatible with the Matbench Discovery benchmark dataset, and does not include any structures that overlap with the WBM dataset. Additionally, the README mentions the sAlex dataset, which is a subsampled version for fine-tuning OMat models. The README also provides instructions for reading data using the fairchem library, and offers support channels for resolving any issues.
Meta Open Materials 2024 (OMat24) Dataset
概述
OMat24数据集包含非平衡结构的单点计算和结构弛豫的混合数据。数据集中的结构标注了总能量（eV）、力（eV/A）和应力（eV/A^3）。数据集以ASE DB兼容的lmdb文件格式提供。
数据集
OMat24数据集
OMat24数据集分为训练集和验证集，每个集包含基于不同输入生成策略的多个子数据集。
训练集
| 子数据集 | 大小 | 下载链接 |
|---|---|---|
| rattled-1000 | 11,388,510 | rattled-1000.tar.gz |
| rattled-1000-subsampled | 3,879,741 | rattled-1000-subsampled.tar.gz |
| rattled-500 | 6,922,197 | rattled-500.tar.gz |
| rattled-500-subsampled | 3,975,416 | rattled-500-subsampled.tar.gz |
| rattled-300 | 6,319,139 | rattled-300.tar.gz |
| rattled-300-subsampled | 3,464,007 | rattled-300-subsampled.tar.gz |
| aimd-from-PBE-1000-npt | 21,269,486 | aimd-from-PBE-1000-npt.tar.gz |
| aimd-from-PBE-1000-nvt | 20,256,650 | aimd-from-PBE-1000-nvt.tar.gz |
| aimd-from-PBE-3000-npt | 6,076,290 | aimd-from-PBE-3000-npt.tar.gz |
| aimd-from-PBE-3000-nvt | 7,839,846 | aimd-from-PBE-3000-nvt.tar.gz |
| rattled-relax | 9,433,303 | rattled-relax.tar.gz |
| 总计 | 100,824,585 | - |
验证集
| 子数据集 | 大小 | 下载链接 |
|---|---|---|
| rattled-1000 | 122,937 | rattled-1000.tar.gz |
| rattled-1000-subsampled | 41,786 | rattled-1000-subsampled.tar.gz |
| rattled-500 | 75,167 | rattled-500.tar.gz |
| rattled-500-subsampled | 43,068 | rattled-500-subsampled.tar.gz |
| rattled-300 | 68,593 | rattled-300.tar.gz |
| rattled-300-subsampled | 37,393 | rattled-300-subsampled.tar.gz |
| aimd-from-PBE-1000-npt | 223,574 | aimd-from-PBE-1000-npt.tar.gz |
| aimd-from-PBE-1000-nvt | 215,589 | aimd-from-PBE-1000-nvt.tar.gz |
| aimd-from-PBE-3000-npt | 65,244 | aimd-from-PBE-3000-npt.tar.gz |
| aimd-from-PBE-3000-nvt | 84,063 | aimd-from-PBE-3000-nvt.tar.gz |
| rattled-relax | 99,968 | rattled-relax.tar.gz |
| 总计 | 1,077,382 | - |
sAlex数据集
sAlex数据集用于OMat模型的微调。sAlex是原始Alexandria数据集的子采样版本，符合Matbench-Discovery标准。
| 数据集 | 分割 | 大小 | 下载链接 |
|---|---|---|---|
| sAlex | 训练集 | 10,447,765 | train.tar.gz |
| sAlex | 验证集 | 553,218 | val.tar.gz |
数据读取
OMat24和sAlex数据集可以通过fairchem库访问。数据集文件以AseLMDBDatabase对象的形式写入，这些对象是ASE Database的LMDB格式实现。
引用
OMat24数据集根据Creative Commons Attribution 4.0 License授权。如果使用此工作，请引用：
@misc{barroso_omat24, title={Open Materials 2024 (OMat24) Inorganic Materials Dataset and Models}, author={Luis Barroso-Luque and Muhammed Shuaibi and Xiang Fu and Brandon M. Wood and Misko Dzamba and Meng Gao and Ammar Rizvi and C. Lawrence Zitnick and Zachary W. Ulissi}, year={2024}, eprint={2410.12771}, archivePrefix={arXiv}, primaryClass={cond-mat.mtrl-sci}, url={https://arxiv.org/abs/2410.12771}, }
