I. M. Sezer
AI Engineer
OrbNet Playground
Drop data orbs, tune activation/depth/width, and train a tiny classifier map with adjustable max steps.
Accuracy 0.0%
Loss 0.000
Score 46
Max Steps 300
Stars ☆☆☆
Challenge Controls
300
Range: 30-3000
Network Structure
L0:2→L1:16→L2:16→L3:1
Hidden layout: 2 -> 16 -> 16 -> 1
Total parameters: 337
How to play
1. Click to place Sky orbs.
2. Drag orbs to reshape the dataset.
3. Right-click orb to flip class.
4. Train to your max-step target and chase 3 stars.
Prediction Map
0/300 steps
Activation: relu
Depth/Width: 2/16
Data Orbs: 80
Challenge: moons