NLG Meta-evaluation on CUS-QA orig. (uk)
0.681Kendall CorrelationQwen 3 30B
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen 3 30BShot=Zero2026.03 | 0.681 | |
| Llama 3.3 70BShot=Few2026.03 | 0.584 | |
| Llama 4 ScoutShot=Zero2026.03 | 0.573 | |
| Llama 4 ScoutShot=Few2026.03 | 0.533 | |
| Qwen 3 30BShot=Few2026.03 | 0.516 | |
| Llama 3.3 70BShot=Zero2026.03 | 0.474 |