Hide & Seek RL LabBack to lab

LAB CONFIGURATION

Lab settings

Recommended values are already applied. Changes are saved in this browser.

Training preset

Applied when the next run starts.

RECOMMENDED4 × 357,258

Four independent Actor-Critic policies, standard size

Default training roundsCan still be adjusted before each run
500
1005,000
Population
4 networks
Parameters per network
357,258
RL algorithm
PPO + GAE
Compute backend
Auto · WebGPU / WebGL

Appearance & language

Follows your device by default.

ThemeUpdates when your system theme changes
LanguageDetected from your browser language
Reduce motionDisables trails and non-essential transitions

Simulation view

Visual only; training is unaffected.

Visibility conesShow each seeker’s field of view
Movement trailsKeep a subtle 18-frame trail

Local checkpoints

Model weights and progress are stored in IndexedDB.

AutosaveSave every 10 rounds and when a run finishes