Confidence Estimation on Global-MMLU English
0.75AUROCLast Answer
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Last AnswerProbe hidden state=last answer token, Training Language=French, Test Language=English, Zero-shot Transfer=true2026.05 | 0.75 | 0.61 | 0.22 | 0.17 | |
| Last QueryProbe hidden state=last query token, Training Language=French, Test Language=English, Zero-shot Transfer=true2026.05 | 0.7 | 0.57 | 0.26 | 0.22 |