Confidence Estimation on Global-MMLU French
0.76AUROCLast Answer
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Last AnswerProbe hidden state=last answer token, Training Language=French, Test Language=French2026.05 | 0.76 | 0.51 | 0.19 | 0.15 | |
| Last QueryProbe hidden state=last query token, Training Language=French, Test Language=French2026.05 | 0.73 | 0.48 | 0.21 | 0.17 |