Adversarial Robustness on Skill-Composed 2k queries (test)
169Explicit Refusals CountRandom
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| RandomBudget=502026.04 | 169 | — | |
| AUTOSKILLBudget=502026.04 | 174 | 5 | |
| RandomBudget=602026.04 | 363 | — | |
| AUTOSKILLBudget=602026.04 | 382 | 19 | |
| RandomBudget=702026.04 | 515 | — | |
| AUTOSKILLBudget=702026.04 | 540 | 25 | |
| RandomBudget=802026.04 | 660 | — | |
| AUTOSKILLBudget=802026.04 | 690 | 30 | |
| RandomBudget=902026.04 | 795 | — | |
| AUTOSKILLBudget=902026.04 | 824 | 29 | |
| RandomBudget=1002026.04 | 905 | — | |
| AUTOSKILLBudget=1002026.04 | 940 | 35 |