Hugging Face's logo

Datasets:
facebook
/
OMAT24

ArXiv:
Tags:
chemistry
materials
License:
Dataset card Data Studio Files
xet
Community
8
mshuaibi commited on
Commit
6ad1078
·
verified ·
1 Parent(s): 91119fc

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +124 -3
README.md CHANGED
@@ -1,3 +1,124 @@
1
- ---
2
- license: cc-by-4.0
3
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: cc-by-4.0
3
+ tags:
4
+ - chemistry
5
+ - materials
6
+ ---
7
+
8
+ <h1 align="center" style="font-size: 36px;">Meta Open Materials 2024 (OMat24) Dataset</h1>
9
+
10
+ <p align="center">
11
+ <img width="559" height="200" src="https://cdn-uploads.huggingface.co/production/uploads/67004f02d66ad0efb0d494c3/yYySyR4CZjnRr09MB33bS.png"?
12
+ </p>
13
+
14
+ ## Overview
15
+ Several datasets were utilized in this work. We provide open access to all datasets used to help accelerate research in the community.
16
+ This includes the OMat24 dataset as well as our modified sAlex dataset. Details on the different datasets are provided below.
17
+
18
+ ## Datasets
19
+
20
+ ### OMat24 Dataset
21
+ The OMat24 dataset contains a mix of single point calculations of non-equilibrium structures and
22
+ structural relaxations. The dataset contains structures labeled with total energy (eV), forces (eV/A)
23
+ and stress (eV/A^3). The dataset is provided in ASE DB compatible lmdb files.
24
+
25
+ The OMat24 train and val splits are fully compatible with the Matbench Discovery benchmark test set.
26
+ 1. The splits do not contain any structure that has a protostructure label present in the initial or relaxed
27
+ structures of the WBM dataset.
28
+ 2. The splits do not include any structure that was generated starting from an Alexandria relaxed structure with
29
+ protostructure lable in the intitial or relaxed structures of the WBM datset.
30
+
31
+ We provide two splits - train and val. Each split is comprised of several subdatasets based on the different input generation strategies, see paper for more details.
32
+ ##### Train
33
+ | Sub-dataset | Size | Download |
34
+ | :-----: | :--: | :------: |
35
+ | rattled-1000 | 11,388,510 | [rattled-1000.tar.gz](https://dl.fbaipublicfiles.com/opencatalystproject/data/omat/241018/omat/train/rattled-1000.tar.gz) |
36
+ | rattled-1000-subsampled | 3,879,741 | [rattled-1000-subsampled.tar.gz](https://dl.fbaipublicfiles.com/opencatalystproject/data/omat/241018/omat/train/rattled-1000-subsampled.tar.gz) |
37
+ | rattled-500 | 6,922,197 | [rattled-500.tar.gz](https://dl.fbaipublicfiles.com/opencatalystproject/data/omat/241018/omat/train/rattled-500.tar.gz) |
38
+ | rattled-500-subsampled | 3,975,416 | [rattled-500-subsampled.tar.gz](https://dl.fbaipublicfiles.com/opencatalystproject/data/omat/241018/omat/train/rattled-500-subsampled.tar.gz) |
39
+ | rattled-300 | 6,319,139 | [rattled-300.tar.gz](https://dl.fbaipublicfiles.com/opencatalystproject/data/omat/241018/omat/train/rattled-300.tar.gz) |
40
+ | rattled-300-subsampled | 3,464,007 | [rattled-300-subsampled.tar.gz](https://dl.fbaipublicfiles.com/opencatalystproject/data/omat/241018/omat/train/rattled-300-subsampled.tar.gz) |
41
+ | aimd-from-PBE-1000-npt | 21,269,486 | [aimd-from-PBE-1000-npt.tar.gz](https://dl.fbaipublicfiles.com/opencatalystproject/data/omat/241018/omat/train/aimd-from-PBE-1000-npt.tar.gz) |
42
+ | aimd-from-PBE-1000-nvt | 20,256,650 | [aimd-from-PBE-1000-nvt.tar.gz](https://dl.fbaipublicfiles.com/opencatalystproject/data/omat/241018/omat/train/aimd-from-PBE-1000-nvt.tar.gz) |
43
+ | aimd-from-PBE-3000-npt | 6,076,290 | [aimd-from-PBE-3000-npt.tar.gz](https://dl.fbaipublicfiles.com/opencatalystproject/data/omat/241018/omat/train/aimd-from-PBE-3000-npt.tar.gz) |
44
+ | aimd-from-PBE-3000-nvt | 7,839,846 | [aimd-from-PBE-3000-nvt.tar.gz](https://dl.fbaipublicfiles.com/opencatalystproject/data/omat/241018/omat/train/aimd-from-PBE-3000-nvt.tar.gz) |
45
+ | rattled-relax | 9,433,303 | [rattled-relax.tar.gz](https://dl.fbaipublicfiles.com/opencatalystproject/data/omat/241018/omat/train/rattled-relax.tar.gz) |
46
+ | Total | 100,824,585 | - |
47
+
48
+ ##### Val
49
+ | Sub-dataset | Size | Download |
50
+ | :-----: | :--: | :------: |
51
+ | rattled-1000 | TODO | TODO |
52
+ | rattled-1000-subsampled | TODO | TODO |
53
+ | rattled-500 | TODO | TODO |
54
+ | rattled-500-subsampled | TODO | TODO |
55
+ | rattled-300 | TODO | TODO |
56
+ | rattled-300-subsampled | TODO | TODO |
57
+ | aimd-from-PBE-1000-npt | TODO | TODO |
58
+ | aimd-from-PBE-1000-nvt | TODO | TODO |
59
+ | aimd-from-PBE-3000-npt | TODO | TODO |
60
+ | aimd-from-PBE-3000-nvt | TODO | TODO |
61
+ | rattled-relax | TODO | TODO |
62
+ | Total | TODO | - |
63
+
64
+ ### sAlex Dataset
65
+ We also provide the sAlex dataset used for fine-tuning of our OMat models. sAlex is a subsampled, Matbench-Discovery compliant, version of the original [Alexandria](https://alexandria.icams.rub.de/).
66
+ sAlex was created by removing structures matched in WBM and only sampling structure along a trajectory with an energy difference greater than 10 meV/atom. For full details,
67
+ please see the manuscript.
68
+ | Dataset | Split | Size | Download |
69
+ | :-----: | :---: | :--: | :------: |
70
+ | sAlex | train | 10,447,765 | [train.tar.gz](dl.fbaipublicfiles.com/opencatalystproject/data/omat/241018/sAlex/train.tar.gz) |
71
+ | sAlex | val | 553,218 | [val.tar.gz](dl.fbaipublicfiles.com/opencatalystproject/data/omat/241018/sAlex/val.tar.gz) |
72
+
73
+ ## How to read the data
74
+ The OMat24 and sAlex datasets can be accessed with the [fairchem](https://github.com/FAIR-Chem/fairchem) library. This package can be installed with:
75
+ ```
76
+ pip install fairchem-core
77
+ ```
78
+
79
+ Dataset files are written as `AseLMDBDatabase` objects which are an implementation of an [ASE Database](https://wiki.fysik.dtu.dk/ase/ase/db/db.html),
80
+ in LMDB format. A single **.aselmdb* file can be read and queried like any other ASE DB (not recommended as there are many files!).
81
+
82
+ You can also read many DB files at once and access atoms objects using the `AseDBDataset` class.
83
+
84
+ For example to read the **rattled-relax** subdataset,
85
+ ```python
86
+ from fairchem.core.datasets import AseDBDataset
87
+
88
+ dataset_path = "/path/to/omat24/train/rattled-relax"
89
+ config_kwargs = {} # see tutorial on additional configuration
90
+
91
+ dataset = AseDBDataset(config=dict(src=dataset_path, **config_kwargs))
92
+
93
+ # atoms objects can be retrieved by index
94
+ atoms = dataset.get_atoms(0)
95
+ ```
96
+
97
+ To read more than one subdataset you can simply pass a list of subdataset paths,
98
+ ```python
99
+ from fairchem.core.datasets import AseDBDataset
100
+
101
+ config_kwargs = {} # see tutorial on additional configuration
102
+ dataset_paths = [
103
+ "/path/to/omat24/train/rattled-relax",
104
+ "/path/to/omat24/train/rattled-1000-subsampled",
105
+ "/path/to/omat24/train/rattled-1000"
106
+ ]
107
+ dataset = AseDBDataset(config=dict(src=dataset_paths, **config_kwargs))
108
+ ```
109
+ To read all of the OMat24 training or validations splits simply pass the paths to all subdatasets.
110
+
111
+ ## Citation
112
+
113
+ The OMat24 dataset is licensed under a [Creative Commons Attribution 4.0 License](https://creativecommons.org/licenses/by/4.0/legalcode). If you use this work, please cite:
114
+ ```
115
+ @misc{barroso_omat24,
116
+ title={Open Materials 2024 (OMat24) Inorganic Materials Dataset and Models},
117
+ 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},
118
+ year={2024},
119
+ eprint={2410.12771},
120
+ archivePrefix={arXiv},
121
+ primaryClass={cond-mat.mtrl-sci},
122
+ url={https://arxiv.org/abs/2410.12771},
123
+ }
124
+ ```