Machine Translation on MGSM (Per-Language Performance)
72.89Bn ScoreTAPO-ChrF++
Evaluation Results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| TAPO-ChrF++Base Model=Llama2026.03 | 72.89 | 85.87 | 89.25 | 85.6 | 75.96 | 85.77 | 63.62 | 68.28 | 77.3 | 83.71 | 78.82 | |
| TAPO-XCOMETBase Model=Llama2026.03 | 72.52 | 84.6 | 87.86 | 85.3 | 74.71 | 84.28 | 61.84 | 68.49 | 76.13 | 81.87 | 77.76 | |
| TAPO-CKBase Model=Llama2026.03 | 72.12 | 83.16 | 87.62 | 84.38 | 75.92 | 83.76 | 59.5 | 68 | 76.44 | 81.76 | 77.27 | |
| TAPO-XCOMETBase Model=Qwen2026.03 | 70.88 | 51.49 | 51.52 | 54.85 | 84.47 | 87.07 | 0 | 41.64 | 82.71 | 85.72 | 61.04 | |
| TAPO-ChrF++Base Model=Qwen2026.03 | 68.89 | 87.47 | 89.83 | 87.11 | 83.32 | 88.8 | 29.29 | 42.37 | 82.7 | 86.6 | 74.64 | |
| TAPO-CKBase Model=Qwen2026.03 | 67.06 | 85.82 | 89.02 | 87.21 | 82.48 | 87.51 | 30.03 | 42.9 | 80.96 | 85.67 | 73.87 | |
| GRPO-TransTestBase Model=Llama2026.03 | 66.38 | 78.76 | 81.44 | 81.3 | 69.55 | 78.35 | 54.99 | 62.74 | 70.76 | 75.58 | 71.98 | |
| GRPO-EnCoTBase Model=Qwen2026.03 | 62.97 | 83.58 | 87.88 | 86.66 | 80.26 | 88.88 | 18.18 | 29.31 | 77.99 | 83.66 | 69.94 | |
| GRPO-TransTestBase Model=Qwen2026.03 | 62.59 | 82.69 | 86.76 | 85.89 | 79.91 | 86.85 | 19.94 | 33.67 | 79.09 | 83.87 | 70.12 | |
| LlamaBase Model=Llama2026.03 | 59.59 | 74.88 | 81.14 | 76.92 | 59.9 | 76.75 | 50.17 | 55.55 | 59.44 | 65.67 | 66 | |
| QwenBase Model=Qwen2026.03 | 58.6 | 84.07 | 87.3 | 85.89 | 81.1 | 87.28 | 7.72 | 17.82 | 77.68 | 86.72 | 67.42 | |
| GRPO-EnCoTBase Model=Llama2026.03 | 56.9 | 64.54 | 71.26 | 66.4 | 53.91 | 64.92 | 47.86 | 54.32 | 54.96 | 57.92 | 59.3 |