NLG Meta-evaluation on WMT en-is 24
0.36Kendall CorrelationLlama 3.3 70B
Evaluation Results
| Method | Links | |
|---|---|---|
| Llama 3.3 70BShot=Zero2026.03 | 0.36 | |
| Llama 4 ScoutShot=Zero2026.03 | 0.302 | |
| Llama 3.3 70BShot=Few2026.03 | 0.286 | |
| Qwen 3 30BShot=Few2026.03 | 0.249 | |
| Llama 4 ScoutShot=Few2026.03 | 0.233 | |
| Qwen 3 30BShot=Zero2026.03 | 0.201 |