# Revised MD17 (rMD17) Molecular Dynamics Benchmark, ColabFit Conversion

## Facts

- Provider: Christensen and von Lilienfeld (University of Basel); converted by ColabFit Exchange
- License: CC0-1.0 (https://api.figshare.com/v2/articles/12672038)
- Size: 7 files
- Format: Parquet (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/colabfit/rMD17
- Access: public, readable without an account
- DOI: https://doi.org/10.60732/682fe04a
- Category: Chemistry & Materials
- Source: https://doi.org/10.1088/2632-2153/abba6f
- Page: https://scigantic.com/archive/524e082b-b736-5a7f-9252-193d0d334c8a
- Keywords: Chemistry & Materials, molecular dynamics, benchmark, Parquet, Hugging Face

## About

ColabFit Exchange conversion of the revised MD17 (rMD17) benchmark published by Anders S. Christensen and O. Anatole von Lilienfeld (Machine Learning: Science and Technology 2020). rMD17 recomputes structures from the original MD17 molecular dynamics trajectories with tighter settings, removing the numerical noise found in the MD17 energies and forces. It covers 10 small organic molecules (aspirin, azobenzene, benzene, ethanol, malonaldehyde, naphthalene, paracetamol, salicylic acid, toluene and uracil) with 100,000 structures each, computed in ORCA at the PBE/def2-SVP level of theory. ColabFit lists 999,906 configurations of C, H, N and O with energy and atomic forces, with configuration sets per molecule. Measured 2026-11-13: ds.parquet, a co/ folder of 2 parquet files (1.05 GB), and cs/ and cs_co_map/ folders for the per-molecule sets; repository storage 4.3 GB. Rows follow the ColabFit configuration schema (positions, atomic_numbers, cell, pbc, energy, atomic_forces, cauchy_stress, software, method and per-row metadata) and open with pandas.read_parquet, pyarrow or datasets.load_dataset; ds.parquet holds the dataset summary. rMD17 is the standard benchmark for force-field learning with models such as NequIP, MACE, Allegro, SchNet, PaiNN and GDML. Released CC0 1.0 by its authors; please cite Christensen and von Lilienfeld 2020 and the ColabFit Exchange.

## Load the data

```python
from datasets import load_dataset
ds = load_dataset("colabfit/rMD17")
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/524e082b-b736-5a7f-9252-193d0d334c8a
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/524e082b-b736-5a7f-9252-193d0d334c8a
- Over MCP: `get_schema_card` with id `524e082b-b736-5a7f-9252-193d0d334c8a` 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)
- [Molecular dynamics datasets](https://scigantic.com/topics/molecular-dynamics)
