Uncertainty Quantification on WMT 19
0.608COMET AUCKLE
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| KLEBackbone Model=Gemma-3-4B2026.01 | 0.608 | 0.559 | |
| Structure-Aware Minimum Bayes Risk DecodingBackbone Model=Gemma-3-4B2026.01 | 0.608 | 0.584 | |
| SARBackbone Model=Gemma-3-4B2026.01 | 0.606 | 0.574 | |
| Structure-Aware Minimum Bayes Risk DecodingBackbone Model=Qwen3-4B2026.01 | 0.603 | 0.578 | |
| KLEBackbone Model=Qwen3-30B-A3B2026.01 | 0.599 | 0.537 | |
| Structure-Aware Minimum Bayes Risk DecodingBackbone Model=Qwen3-30B-A3B2026.01 | 0.599 | 0.57 | |
| COCOABackbone Model=Qwen3-30B-A3B2026.01 | 0.597 | 0.556 | |
| Structure-Aware Minimum Bayes Risk DecodingBackbone Model=Gemma-3-12B2026.01 | 0.596 | 0.579 | |
| KLEBackbone Model=Qwen3-4B2026.01 | 0.594 | 0.539 | |
| COCOABackbone Model=Qwen3-4B2026.01 | 0.591 | 0.556 | |
| KLEBackbone Model=Gemma-3-12B2026.01 | 0.59 | 0.539 | |
| COCOABackbone Model=Gemma-3-4B2026.01 | 0.589 | 0.555 | |
| SARBackbone Model=Gemma-3-12B2026.01 | 0.587 | 0.562 | |
| COCOABackbone Model=Gemma-3-12B2026.01 | 0.586 | 0.542 | |
| SARBackbone Model=Qwen3-30B-A3B2026.01 | 0.583 | 0.548 | |
| SARBackbone Model=Qwen3-4B2026.01 | 0.58 | 0.549 | |
| MSPBackbone Model=Qwen3-30B-A3B2026.01 | 0.577 | 0.547 | |
| MSPBackbone Model=Qwen3-4B2026.01 | 0.573 | 0.553 | |
| MSPBackbone Model=Gemma-3-12B2026.01 | 0.572 | 0.532 | |
| MSPBackbone Model=Gemma-3-4B2026.01 | 0.563 | 0.535 | |
| P(True)Backbone Model=Gemma-3-12B2026.01 | 0.557 | 0.495 | |
| SEBackbone Model=Gemma-3-4B2026.01 | 0.543 | 0.531 | |
| P(True)Backbone Model=Gemma-3-4B2026.01 | 0.537 | 0.485 | |
| SEBackbone Model=Qwen3-4B2026.01 | 0.533 | 0.517 | |
| P(True)Backbone Model=Qwen3-4B2026.01 | 0.523 | 0.55 | |
| SEBackbone Model=Gemma-3-12B2026.01 | 0.518 | 0.513 | |
| P(True)Backbone Model=Qwen3-30B-A3B2026.01 | 0.517 | 0.506 | |
| SEBackbone Model=Qwen3-30B-A3B2026.01 | 0.517 | 0.512 |