quintana42/gang-of-four-training-data
收藏Hugging Face2026-01-16 更新2026-03-29 收录
下载链接：
https://hf-mirror.com/datasets/quintana42/gang-of-four-training-data
下载链接
链接失效反馈官方服务：
资源简介：
---
tags:
- gang-of-four
- card-game
- imitation-learning
license: mit
---
# Gang of Four Training Data
Training data for the Gang of Four neural AI, generated from ExpertStrategy self-play.
## Dataset Details
- **Format**: JSONL (one JSON object per line)
- **Size**: ~1M game states
- **Source**: ExpertStrategy vs ExpertStrategy games
## Schema
Each line contains:
```json
{
"state": [328 floats],
"action_mask": [40 floats],
"action_idx": int,
"declared_last_card": bool
}
```
## Usage
```python
from training.dataset import GangOfFourDataset
dataset = GangOfFourDataset("data/training_data.jsonl")
```
## Generation
```bash
python scripts/generate_training_data.py --num-games 100000 --output data/training_data.jsonl
```
See: https://github.com/sboily/gang_of_four
提供机构：
quintana42
