tinyppo-snake · in-browser PPO training

Source: original

tinyppo-snake

Train Watch Weights grid Compare runs → +

Environment Sample

idle

Start training

Training metrics

0s

avg500 0 last 500 episodes

peak 0 best window

roll/s 0 20-step avg

progress 0 0 episodes

pol= — val= — ent= — gn= — kl= —

Scores

Events

Trained-policy roll-outs

0 · trained 0

Resume training Reset envs grid 1×1 2×2 3×3 delay 40 ms

parameter live updates Reset view waiting for first snapshot…

Initializing 3D renderer…

step # —

no tensor loaded

Configure runs

preset 3 seeds @ default LR LR sweep (1e-3 / 3e-3 / 1e-2) 2 seeds Custom Apply preset

Start all runs Stop all + Add run idle

no runs yet