webxos/synthetic_GO_dataset

Synthetic GO Dataset

1. Game ID: go_ml00v6sh_ooin8ui0b
2. Total Frames: 821
3. Board Size: 19×19
4. Export Date: 2026-01-29T23:12:28.195Z
5. Generated by: Monotone GO Pro v2.5

File Structure

dataset.zip/
├── images/                    # PNG images (400×400)
│   └── frame_00001.png       # Board visualization
├── metadata/
│   ├── dataset.csv           # CSV metadata
│   ├── dataset.json          # JSON metadata
│   └── game.sgf              # SGF game record
└── README.md                 # This file

Suitable for:

• Machine learning training
• Computer vision
• Reinforcement learning
• Go AI development

Dataset Description

This dataset provides a synthetic, frame-by-frame record of a single professional-level Go game on a 19x19 board. Generated using Monotone GO Pro v2.5, it includes 821 frames capturing the game's progression from start to finish. Each frame consists of a visual board representation (PNG image) and a structured board state (array), enabling applications in AI, machine learning, and game analysis.

Key Metadata

Version: Go Professional Dataset v2.5 Game ID: go_ml00v6sh_ooin8ui0b Total Frames: 821 Board Size: 19x19 Export Date: 2026-01-29T23:12:28.195Z Generated By: Monotone GO Pro v2.5 License: MIT Size: ~15 MB (ZIP), ~14.9 MB (Parquet)

How the Dataset Was Generated

Simulated using Monotone GO Pro v2.5, a Go engine mimicking professional play. The game starts empty and progresses with alternating moves, capturing real-time board evolutions. No human intervention; fully synthetic for reproducibility. Use Cases

-Machine Learning Training: Train models to predict next moves, evaluate board positions, or classify game phases (opening, mid-game, endgame).

-Computer Vision Tasks: Use images for object detection (stones), segmentation (board regions), or generation (e.g., GANs for synthetic boards).

-Reinforcement Learning: Feed board states into RL agents like AlphaGo-style systems for policy/value network training.

-Go AI Development: Benchmark algorithms on sequential data; simulate variations from SGF.

-Strategy Analysis: Visualize move patterns, compute liberties/territories with board arrays.

-Educational Tools: Build apps/tutorials showing game progression; interactive replays.

-Game Simulation: Extend to multi-game datasets; test Monte Carlo Tree Search (MCTS).

-Research: Study complexity in sequential decision-making; compare synthetic vs. real pro games.

Citation

If using, cite as: text@dataset{synthetic_GO_dataset, author = {webxos}, title = {Synthetic GO Dataset}, year = {2026}, url = {webxos.netlify.app} }