vlabki/rr-speed-item-v1
Self-contained recurrent player-policy checkpoint.
- Source directory:
best_reliable - PPO update: 6000
- Environment steps: 36864000
- Action support: bc
Evaluation values below come from deterministic fixed-seed argmax runs. Frame averages exclude DNF races.
Best reliable evaluation
| Metric | Value |
|---|---|
| Checkpoint update | 6000 |
| Cohort | 20 |
| Finish rate | 90.0% |
| Finished mean frames | 11407.17 |
| Finished median frames | 11342.50 |
| Fastest finish frames | 10685 |
| Finished P90 frames | 11994.30 |
| First-place rate among finishes | 100.0% |
| Mean wall events | 0.10 |
| Mean respawns | 0.30 |
Best fastest evaluation
| Metric | Value |
|---|---|
| Checkpoint update | 4500 |
| Cohort | 20 |
| Finish rate | 75.0% |
| Finished mean frames | 11576.27 |
| Finished median frames | 11589.00 |
| Fastest finish frames | 10681 |
| Finished P90 frames | 12030.40 |
| First-place rate among finishes | 100.0% |
| Mean wall events | 0.30 |
| Mean respawns | 0.55 |
Files
Runtime weights, model config, normalization statistics, route reference, training config, and portable checksummed fallbacks are included. Raw rollout traces, optimizer state, and full training logs are excluded.