Baby AlphaGo: Snaky Hexomino

Interactive Deep Reinforcement Learning visualizer for bounding the Snaky game tree on an 8x8 grid.

Current Turn: Maker

MCTS & Neural Network

About this Architecture

This agent is trained using a ResNet architecture with Dual Heads (Policy & Value) combined with Monte Carlo Tree Search (PUCT).

The grid is bounded to 8x8 to allow for rapid self-play iteration while encapsulating the full mathematical reach of the Snaky hexomino.

Inspect SnakeyNet Architecture▶