# Transition1x Training Split: Reaction Pathway DFT Energies and Forces, ColabFit Conversion

## Facts

- Provider: Mathias Schreiner, Arghya Bhowmik, Tejs Vegge, Jonas Busk, Ole Winther (DTU); ColabFit conversion
- License: MIT (https://api.figshare.com/v2/articles/19614657)
- Size: 4 files
- Format: Parquet (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/colabfit/Transition1x_train
- Access: public, readable without an account
- DOI: https://doi.org/10.60732/b1104cc5
- Category: Chemistry & Materials
- Source: https://doi.org/10.1038/s41597-022-01870-w
- Page: https://scigantic.com/archive/36d77193-b275-5862-97fe-ce85fc83fd5c
- Keywords: Chemistry & Materials, DFT, benchmark, Parquet, Hugging Face

## About

The training split of Transition1x, a benchmark of DFT energies and forces for molecular configurations on and around reaction pathways, computed at the wB97x/6-31G(d) level from nudged elastic band (NEB) calculations on about 10,000 organic reactions. The publishers describe 9.6 million DFT calculations in the full release and show that the data represent transition state regions better than equilibrium-focused sets such as QM9 and ANI-1x, which matters for reactive machine learning potentials and barrier height prediction. ColabFit's card for this split counts 62,988 unique configurations and 535,993 atoms (C, H, N, O); the repository is about 0.11 GB (measured 2026-11-14). Validation and test splits are separate ColabFit repositories. 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. The upstream figshare release is MIT licensed; cite Schreiner et al., Scientific Data 2022.

## Load the data

```python
from datasets import load_dataset
ds = load_dataset("colabfit/Transition1x_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/36d77193-b275-5862-97fe-ce85fc83fd5c
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/36d77193-b275-5862-97fe-ce85fc83fd5c
- Over MCP: `get_schema_card` with id `36d77193-b275-5862-97fe-ce85fc83fd5c` 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)
