Confidence Estimation on MKQA (test)
0.82AUROCLast Answer
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Last AnswerLanguage=fr, Protocol=Trained and tested on source language, Backbone=Qwen 3 8B2026.05 | 0.82 | 0.49 | 0.14 | 0.12 | |
| Last AnswerLanguage=es, Protocol=Zero-shot, Backbone=Qwen 3 8B2026.05 | 0.82 | 0.53 | 0.19 | 0.2 | |
| Last AnswerLanguage=ru, Protocol=Zero-shot, Backbone=Qwen 3 8B2026.05 | 0.8 | 0.43 | 0.15 | 0.16 | |
| Last QueryLanguage=fr, Protocol=Trained and tested on source language, Backbone=Qwen 3 8B2026.05 | 0.73 | 0.36 | 0.17 | 0.14 | |
| Last AnswerLanguage=pl, Protocol=Zero-shot, Backbone=Qwen 3 8B2026.05 | 0.73 | 0.35 | 0.23 | 0.24 | |
| Last QueryLanguage=ru, Protocol=Zero-shot, Backbone=Qwen 3 8B2026.05 | 0.72 | 0.3 | 0.12 | 0.11 | |
| Last QueryLanguage=es, Protocol=Zero-shot, Backbone=Qwen 3 8B2026.05 | 0.71 | 0.34 | 0.17 | 0.14 | |
| Last QueryLanguage=en, Protocol=Zero-shot, Backbone=Qwen 3 8B2026.05 | 0.7 | 0.47 | 0.22 | 0.18 | |
| Last AnswerLanguage=en, Protocol=Zero-shot, Backbone=Qwen 3 8B2026.05 | 0.7 | 0.44 | 0.24 | 0.22 | |
| Last QueryLanguage=pl, Protocol=Zero-shot, Backbone=Qwen 3 8B2026.05 | 0.67 | 0.28 | 0.15 | 0.13 | |
| Last AnswerLanguage=ja, Protocol=Zero-shot, Backbone=Qwen 3 8B2026.05 | 0.64 | 0.23 | 0.16 | 0.16 | |
| Last QueryLanguage=ja, Protocol=Zero-shot, Backbone=Qwen 3 8B2026.05 | 0.57 | 0.16 | 0.13 | 0.11 | |
| MajorityLanguage=all*, Backbone=Qwen 3 8B2026.05 | 0.5 | 0.58 | 0.17 | 0.17 | |
| Prior prob.Language=all*, Backbone=Qwen 3 8B2026.05 | 0.5 | 0.58 | 0.14 | 0.04 |