Runtime Controllability on CoSApien (test)
100Allow: G+/STR1S-8b/pos
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| TR1S-8b/posParameters=8B, Variant=pos2026.02 | 100 | 88.7 | 88.8 | 28.1 | 68.5 | |
| PACT2026.02 | 83.3 | 95.9 | 93.1 | 15 | 68 | |
| gpt-oss-20bParameters=20B2026.02 | 68 | 84.7 | 91.7 | 2.5 | 59.6 | |
| gpt-oss-120bParameters=120B2026.02 | 66.7 | 97.4 | 96.9 | 0 | 64.8 | |
| TR1S-8b/adhParameters=8B, Variant=adh2026.02 | 58.3 | 95.2 | 91 | 10.3 | 65.5 |