Home▸AI & Machine Learning▸Alignment Methods Overview: 2D Reward Map

🧭 Alignment Methods Overview: 2D Reward Map

Top-down reward map where a policy is trained live with SFT, RLHF, DPO, ORPO or RLAIF update rules.

AI & Machine Learning2DModerate60 FPS📱 Mobile-adapted⇄ 3D version

⚙ Under the hood

Top-down reward map where a policy is trained live with SFT, RLHF, DPO, ORPO or RLAIF update rules.

alignmentRLHFDPO

2D · HTML5 Canvas 2D · 60 FPS target · runs fully client-side, no install