# JARVIS-ML CFID Descriptors for 460K OQMD Crystals: DFT Formation Energies (NIST JARVIS, figshare), ColabFit Conversion

## Facts

- Provider: OQMD (Kirklin, Saal, Wolverton et al.) and Kamal Choudhary (NIST JARVIS); ColabFit conversion
- License: CC-BY-4.0 (https://api.figshare.com/v2/articles/13055333)
- Size: 5 files
- Format: Parquet (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/colabfit/JARVIS_CFID_OQMD
- Access: public, readable without an account
- DOI: https://doi.org/10.6084/m9.figshare.13055333
- Category: Chemistry & Materials
- Source: https://doi.org/10.1038/npjcompumats.2015.10
- Page: https://scigantic.com/archive/d881a2e7-6e91-5d33-9505-7b6416054e31
- Keywords: Chemistry & Materials, DFT, benchmark, Parquet, Hugging Face

## About

Crystal structures and DFT formation energies for about 460,000 inorganic compounds from the Open Quantum Materials Database (OQMD, computed with VASP using the PAW-PBE functional), packaged by NIST's JARVIS team with classical force-field inspired descriptors (CFID) for every entry. CFID encodes each crystal as a fixed-length feature vector, so the set can be used directly to train and benchmark machine-learning models that predict formation energy and thermodynamic stability of inorganic compounds. ColabFit's card counts 459,943 configurations and about 2.37 million atoms over 89 elements, labelled with formation energy; the repository is about 12.9 GB. Files are the ColabFit Exchange conversion: Parquet files under co/ hold one row per configuration (atomic numbers, positions, cell, computed properties and metadata), ds.parquet holds dataset-level information, and cs/ groups configurations into named sets where defined; read them with pyarrow, pandas, polars or Hugging Face datasets. The JARVIS-ML-CFID-OQMD figshare deposit (DOI 10.6084/m9.figshare.13055333) is released under CC BY 4.0. Cite Kirklin et al., npj Computational Materials 1, 15010 (2015) for OQMD and Choudhary et al., Physical Review Materials 2, 083801 (2018) for CFID.

## Load the data

```python
from datasets import load_dataset
ds = load_dataset("colabfit/JARVIS_CFID_OQMD")
print(ds)
```

## Data structure

File types in sampled files: Parquet

Example file: `co/co_0.parquet`

## Open it

- In a hosted notebook with the data mounted: https://scigantic.com/archive/d881a2e7-6e91-5d33-9505-7b6416054e31
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/d881a2e7-6e91-5d33-9505-7b6416054e31
- Over MCP: `get_schema_card` with id `d881a2e7-6e91-5d33-9505-7b6416054e31` at https://api.scigantic.com/mcp

## Related topics

- [Parquet datasets](https://scigantic.com/topics/parquet)
- [Benchmark datasets](https://scigantic.com/topics/benchmark)
- [Chemistry & Materials datasets](https://scigantic.com/topics/chemistry-and-materials)
- [DFT datasets](https://scigantic.com/topics/dft)
