# Open Molecules 2025 (OMol25) Full Training Set, ColabFit Conversion

## Facts

- Provider: Meta FAIR Chemistry and Lawrence Berkeley National Laboratory (Levine et al.); ColabFit conversion
- License: CC-BY-4.0 (https://raw.githubusercontent.com/facebookresearch/fairchem/main/docs/molecules/datasets/omol25.md)
- Size: 456 files
- Format: Parquet (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/colabfit/OMol25_train
- Access: public, readable without an account
- DOI: https://doi.org/10.60732/41666b82
- Category: Chemistry & Materials
- Source: https://arxiv.org/abs/2505.08762
- Page: https://scigantic.com/archive/84ec5637-91e4-5fe7-826b-2f7faa9d84df
- Keywords: Chemistry & Materials, DFT, interatomic potentials, Parquet, Hugging Face

## About

The full training split of Open Molecules 2025 (OMol25), the molecular DFT dataset released by Meta FAIR Chemistry with Lawrence Berkeley National Laboratory and collaborators. Every calculation is at the range-separated meta-GGA wB97M-V/def2-TZVPD level of theory in ORCA 6, covering biomolecules (protein pockets, protein-ligand and nucleic acid fragments), metal complexes and organometallics, electrolytes and ionic liquids, reactive systems, and recomputed community sets such as SPICE, ANI-2x, Transition1x, GEOM and OrbNet Denali, with charges and spin multiplicities across the periodic table. ColabFit's card counts 101,666,280 configurations, about 5.24 billion atoms and 83 elements, labelled with total energy and atomic forces; the repository holds 453 Parquet shards totalling about 306 GB (measured 2026-11-15). OMol25 is the training data for the UMA and eSEN universal machine learning interatomic potentials and for molecular MLIP benchmarks. Meta's own Hugging Face copy is behind an access form; this ColabFit copy is anonymously readable. This is the ColabFit Exchange conversion: configurations are Parquet files under co/ (structure, positions, cell, computed properties and metadata per row) with dataset-level information in ds.parquet; read them with pyarrow, pandas, polars or Hugging Face datasets, following the ColabFit configuration schema at materials.colabfit.org/docs. OMol25 is licensed by Meta under CC-BY-4.0; cite Levine et al., arXiv:2505.08762 (2025).

## Load the data

```python
from datasets import load_dataset
ds = load_dataset("colabfit/OMol25_train")
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/84ec5637-91e4-5fe7-826b-2f7faa9d84df
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/84ec5637-91e4-5fe7-826b-2f7faa9d84df
- Over MCP: `get_schema_card` with id `84ec5637-91e4-5fe7-826b-2f7faa9d84df` at https://api.scigantic.com/mcp

## Related topics

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