Run configuration
Hyperparameters
Outcomes
Reward per episode
Each point is one episode — green finished a lap, orange did not. The line is a rolling mean over one iteration.
Reward by iteration
Mean with min–max spread.
Track completion
Mean progress and share of episodes finishing a lap.
Lap times
Completed laps only.
Driven paths
Track heatmap
Filtered steps binned over the track surface.
Episode replay
Pick an episode to inspect step by step.
—