bridge-dds-dataset / metadata.jsonl
wopdevries's picture
Update metadata.jsonl
66fe9b7 verified
# Bridge DDS Dataset
This dataset contains bridge deals in Portable Bridge Notation (PBN) format with Double Dummy Solver (DDS) results, generated using [make-dds-dataset](https://github.com/wopdevries/make-dds-dataset). It is designed for reinforcement learning (RL) research in bridge, offering reproducible, ML-ready data.
## Dataset Description
- **Homepage**: [https://github.com/wopdevries/make-dds-dataset](https://github.com/wopdevries/make-dds-dataset)
- **Repository**: [https://github.com/wopdevries/make-dds-dataset](https://github.com/wopdevries/make-dds-dataset)
- **Paper**: None
- **Point of Contact**: Open an issue on the GitHub repository
- **Size**: ~11,293 bytes for 100 deals
- **Format**: TSV with PBN strings (e.g., `N:A2.KT.T763.AKQ87 ...`) and DDS results (20 comma-separated integers for tricks by player and denomination)
- **License**: GPL v3 (code), Apache 2.0 (`libdds.so`)
## Dataset Structure
- **Files**:
- `sample_results.tsv`: TSV file with 100 deals, each with a PBN string and DDS results.
- Columns:
- `pbn`: String representing the deal in PBN format.
- `dds_results`: Sequence of 20 integers (tricks for N, E, S, W in C, D, H, S, NT).
- **Splits**: Single split (`train`) with 100 examples.
- **Example**:
```python
import pandas as pd
data = pd.read_csv('sample_results.tsv', sep='\t')
print(data.head())