Machine Translation on Xibe sjo (test)
10.81BLEUgemma-4-31B-it
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| gemma-4-31B-itReasoning=true2026.06 | 10.81 | 37.36 | 64.28 | 63.29 | |
| gemma-4-31B-itReasoning=false2026.06 | 9.84 | 35.85 | 62.31 | 59.69 | |
| Qwen3-14BReasoning=true2026.06 | 4.7 | 30.83 | 57.09 | 48.74 | |
| Qwen3-14BReasoning=false2026.06 | 3.11 | 28.99 | 52.79 | 43.56 | |
| Qwen3-8BReasoning=true2026.06 | 2.74 | 27.79 | 55.57 | 43.23 | |
| Qwen3-8BReasoning=false2026.06 | 2.08 | 26.69 | 48.43 | 35.97 | |
| gemma-4-E4B-itReasoning=true2026.06 | 1.88 | 22.7 | 54.49 | 45.38 | |
| Qwen3-4B-ThinkingReasoning=true2026.06 | 1.21 | 17.54 | 53.21 | 41.82 | |
| gemma-4-E2B-itReasoning=false2026.06 | 0.76 | 22.15 | 47.57 | 42.87 | |
| gemma-4-E4B-itReasoning=false2026.06 | 0.67 | 16.05 | 49.35 | 40.01 | |
| gemma-4-E2B-itReasoning=true2026.06 | 0.58 | 17.49 | 51.3 | 39.48 | |
| Qwen3-4B-ThinkingReasoning=false2026.06 | 0.18 | 6.71 | 35.72 | 28.6 |