Tool-use Policy Auditing on τ2-BENCH Telecom base
20.2PASS^4PG-CHECKLIST
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| PG-CHECKLISTVariant=CHECKLIST, Agent and Verifier Model=GPT 5.4, n=4, Verifier type=paired-verifier2026.06 | 20.2 | 10.2 | 53.8 | |
| BaselineAgent and Verifier Model=GPT 5.4, n=4, Verifier type=paired-verifier2026.06 | 19.3 | 9.1 | 53.8 | |
| PG-RAWVariant=RAW, Agent and Verifier Model=GPT 5.4, n=4, Verifier type=paired-verifier2026.06 | 14.9 | 6.8 | 42.3 |