Machine Translation on Reference-based Translation Dataset Mongolian-to-Russian
9.04BLEU-3CoPiT
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| CoPiTBackbone=Ministral-3 (14B), Evaluation Protocol=Fine-tuned2026.07 | 9.04 | 5.84 | 32.49 | 29.12 | 72.8 | |
| CoPiTBackbone=GPT-4.1, Evaluation Protocol=Zero-shot2026.07 | 8.71 | 5.63 | 31.84 | 28.53 | 69.4 | |
| CoPiTBackbone=Qwen-3 (4B), Evaluation Protocol=Fine-tuned2026.07 | 3.45 | 1.99 | 18.6 | 16.24 | 54.4 | |
| CoPiTBackbone=Qwen-3 (30B), Evaluation Protocol=Fine-tuned2026.07 | 2.83 | 1.55 | 19.51 | 17 | 48.7 | |
| CoPiTBackbone=Ministral-3 (14B), Evaluation Protocol=Zero-shot2026.07 | 2.31 | 1.12 | 21.7 | 19.05 | 55.8 | |
| CoPiTBackbone=Ministral-3 (3B), Evaluation Protocol=Fine-tuned2026.07 | 2.21 | 1 | 21.92 | 19.08 | 54.7 | |
| DirectBackbone=GPT-4.1, Evaluation Protocol=Direct2026.07 | 1.57 | 0.73 | 15.86 | 14 | 51.6 | |
| DirectBackbone=Ministral-3 (14B), Evaluation Protocol=Direct2026.07 | 0.92 | 0.34 | 16.7 | 14.44 | 47.3 | |
| CoPiTBackbone=Qwen-3 (30B), Evaluation Protocol=Zero-shot2026.07 | 0.83 | 0.39 | 13.41 | 11.54 | 38.8 | |
| CoPiTBackbone=Ministral-3 (3B), Evaluation Protocol=Zero-shot2026.07 | 0.69 | 0.26 | 16.34 | 14.15 | 43.9 | |
| DirectBackbone=Ministral-3 (3B), Evaluation Protocol=Direct2026.07 | 0.6 | 0.2 | 14.08 | 12.03 | 40.4 | |
| CoPiTBackbone=Qwen-3 (4B), Evaluation Protocol=Zero-shot2026.07 | 0.59 | 0.31 | 11.42 | 9.37 | 31.2 | |
| DirectBackbone=Qwen-3 (4B), Evaluation Protocol=Direct2026.07 | 0.35 | 0.1 | 5.03 | 4.65 | 26.6 | |
| DirectBackbone=Qwen-3 (30B), Evaluation Protocol=Direct2026.07 | 0.33 | 0.12 | 9.55 | 8.33 | 33.1 |