Safety Gate Evaluation on Safety Self-Improvement Testbeds
500Sum TPR (N=500)Ball verifier
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Ball verifierδ=0, Mutation scale=σ∗ ∝ d−0.54, Compute=O(d)2026.03 | 500 | — |
| Method | Links | ||
|---|---|---|---|
| Ball verifierδ=0, Mutation scale=σ∗ ∝ d−0.54, Compute=O(d)2026.03 | 500 | — |