a1exmah0ne/othello-ai-model

Othello AI Model (Auto-Updated)

This model is automatically improved by playing games on the Othello website!

Model Details

• Format: ONNX (Open Neural Network Exchange)
• Architecture: ResNet with 19 residual blocks, 256 channels
• Input: 3x8x8 tensor (empty, black, white squares)
• Outputs:
  • Policy: 64 move logits
  • Value: Position evaluation (-1 to +1)
• Size: ~96 KB (optimized for web)

How It Works

1. Players play games on the Othello website
2. Game data is collected
3. Model trains on this data weekly
4. Improved model is automatically uploaded here
5. Website downloads new model on next refresh

Updates

• Last updated: 2025-11-07 14:56:59 UTC
• Training samples: Generated from collected gameplay
• Improvement: Model learns from real games!

Usage

import onnxruntime as ort
import numpy as np

session = ort.InferenceSession('othello_model.onnx')
board = np.zeros((1, 3, 8, 8), dtype=np.float32)
outputs = session.run(None, {'input': board})
policy = outputs[0]  # Move logits
value = outputs[1]   # Position evaluation

---

Auto-updated by Othello Continuous Learning Pipeline