🏆 Threes! AI: deck-aware expectimax, N-tuple TD learning, and AlphaZero
-
Updated
Aug 24, 2026 - Python
🏆 Threes! AI: deck-aware expectimax, N-tuple TD learning, and AlphaZero
Browser-based Threes AI. 39.1% reach 6144 in 1,024 games. TD learning + five-ply expectimax. English and Chinese. MIT.
Computer game AI — Threes! game agent with N-tuple network
Threes-style sliding merge puzzle in Flutter — each tile moves at most one cell per swipe (not a full 2048-style slide), with the signature 1+2=3 special-pair merge rule.
smaller than threes: https://jamiebayne.co.uk/games/ones
A Python console-based Threes puzzle simulator with move handling, tile merging, scoring, and simple solver strategies.
CITS3001 AI project
Set of exercises to practice data structure in C and Java.
A customised Open-AI gym environment which simulate the puzzle video game Threes
QLearning for the Threes game
To associate your repository with the threes topic, visit your repo's landing page and select "manage topics."