Λ Cadence

Pragma Research · Repair networks · AI

Cadence

Brains that answer by settling. Patches wired both ways repair their disagreement until the whole brain agrees — then remember, adapt, and repair locally from what they witness.

Early research · every demo runs on cadence-net 0.95.2

Tron

[model card]

Light cycles against a brain that sees the whole field

> run in browser

Connect Four

[brain training]

A deep brain that judges every position beside its move

> run in browser

Atari Arcade

[model card]

Brains that learn Atari games from pixels — teacher first, then reward

> run in browser

Robot Arena

[brain training]

The Cadence Showcase League — robots raised in a nursery, fighting live

> run in browser

Robot Arm

[model card]

Pick and place under contact physics — move the bin and it follows

> run in browser

Eyes

[model card]

One brain, an eye on every shape — from the page's pixels only

> run in browser

Connectome

[model card]

Teach a fish gaze circuit on measured wiring; explore a larva

> run in browser

Castlevania

[model card]

A brain plays the NES game from what its eye sees on screen

> runs locally