Skip to content

Latest commit

 

History

History
107 lines (78 loc) · 2.85 KB

README.md

File metadata and controls

107 lines (78 loc) · 2.85 KB

Chess ML with Reasoning

A chess AI system that combines deep learning with reasoning capabilities, including causal understanding, strategic planning, and natural language explanation of its decisions.

Features

  • Reasoning System

    • Causal model for understanding move consequences
    • Strategic planning with multi-step reasoning
    • Natural language explanations of decisions
    • Concept learning from gameplay patterns

  • Neural Network Architecture

    • GFlowNet-based reasoning engine
    • Integrated world model for position understanding
    • Concept learner for pattern recognition
    • Inference machine for decision making

  • Training & Evaluation

    • Self-play training with continuous improvement
    • Comprehensive evaluation metrics
    • Model checkpointing and versioning
    • Performance analysis tools

Project Structure

chess-reasoning/
├── src/
│   ├── concepts/        # Chess concept learning
│   ├── inference/       # Decision inference engine
│   ├── language/        # Natural language processing
│   ├── planning/        # Strategic planning
│   ├── reasoning/       # Core reasoning modules
│   ├── world_model/     # Position understanding
│   └── utils/          # Helper utilities
├── scripts/
│   ├── train.py        # Training script
│   └── evaluate.py     # Evaluation script
└── gui/
    └── chess_gui.py    # Interactive interface

Requirements

  • Python 3.8+
  • PyTorch
  • Additional dependencies in setup.py

Installation

# Clone the repository
git clone https://github.com/yourusername/chess-reasoning.git
cd chess-reasoning

# Install the package and dependencies
pip install -e .

Usage

Training the Model

python scripts/train.py --epochs 100 --batch-size 32

Key training parameters:

  • --epochs: Number of training epochs
  • --batch-size: Batch size for training
  • --learning-rate: Learning rate for optimization
  • --checkpoint-dir: Directory for saving model checkpoints

Evaluation

python scripts/evaluate.py --model-path checkpoints/latest.pth

Playing Against the AI

python gui/chess_gui.py

Model Components

1. Causal Model

The system uses a causal model to understand the relationships between moves and their consequences, enabling better strategic planning.

2. Strategic Planner

Implements multi-step reasoning to develop and execute complex strategies during gameplay.

3. Language Explainer

Provides natural language explanations for the AI's decisions, making the system more interpretable and educational.

4. Concept Learner

Automatically learns and recognizes important chess patterns and concepts from gameplay data.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Last Updated

December 28, 2024