| # 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()) |
