README.md

---
license: mit
task_categories:
- tabular-classification
- text-generation
pretty_name: MOFFlow-2 Preprocessed Data
---

# MOFFlow-2 Preprocessed Dataset

Preprocessed data artifacts for **MOFFlow-2** tasks:
- Structure prediction (`csp`)
- Generation (`gen`)

Included assets: LMDB datasets, sequence data (`mof_sequence_*.json`), split files, and metal library (`metal_lib_train.pkl`).

Project repo: https://github.com/nayoung10/MOFFlow-2

## Files

- `lmdb/csp/train.tar.gz`
- `lmdb/csp/val.tar.gz`
- `lmdb/csp/test.tar.gz`
- `lmdb/gen/train.tar.gz`
- `lmdb/gen/val.tar.gz`
- `seqs/seqs_data.tar.gz`
- `splits.tar.gz`
- `metals/metal_lib_train.pkl`
- `SHA256SUMS`

## Quick Start (download + verify + extract)

```bash
# Download all files from this dataset repo
hf download lkny123/mofflow2-data --repo-type dataset --local-dir .

# Verify integrity
sha256sum -c SHA256SUMS
```

Extract into MOFFlow-2 project layout (`$PROJECT_ROOT/data`):

```bash
# Example assumes you are in your MOFFlow-2 root
mkdir -p data/lmdb/csp data/lmdb/gen data/seqs data/splits data/metals/gen

tar -xzf lmdb/csp/train.tar.gz -C data/lmdb/csp
tar -xzf lmdb/csp/val.tar.gz -C data/lmdb/csp
tar -xzf lmdb/csp/test.tar.gz -C data/lmdb/csp

tar -xzf lmdb/gen/train.tar.gz -C data/lmdb/gen
tar -xzf lmdb/gen/val.tar.gz -C data/lmdb/gen

tar -xzf seqs/seqs_data.tar.gz -C data
cp seqs/vocab.json data/seqs/vocab.json
tar -xzf splits.tar.gz -C data
cp metals/metal_lib_train.pkl data/metals/gen/metal_lib_train.pkl
```

Expected result:
- `data/lmdb/csp/MetalOxo_final_{train,val,test}.lmdb`
- `data/lmdb/gen/MetalOxo_final_{train,val}.lmdb`
- `data/seqs/mof_sequence_{train,val}.json`, `data/seqs/vocab.json`
- `data/splits/...`
- `data/metals/gen/metal_lib_train.pkl`

## Citation

If you use this dataset, please cite the MOFFlow-2 paper and repository.