HEX0N
Devlog
- October 03, 2026 by Yotori Kiba#artificial-intelligence, #machine-learning, #reinforcement-learningxen0n started as an attempt to get beyond HEX0N’s original hand-written naive bot. Model-based reinforcement learning (RL) training followed an AlphaZero-style loop using Gumbel search for 2M total... Continue reading

- October 03, 2026 by Yotori Kiba1#artificial-intelligence, #machine-learning, #reinforcement-learningHEX0N update: Normal and Turbo AI modes There's a new Mode choice for the AI opponents. In Normal, both bots play a looser, more forgiving game. Technically, each bot picks its move with probability p... Continue reading
