Machine Translation on WMT19
0.346COMET ScoreSAR
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| SARBackbone=Gemma-3-4B2026.01 | 0.346 | 0.2 | — | — | |
| R_BayesBackbone=Gemma-3-4B2026.01 | 0.333 | 0.224 | — | — | |
| KLEBackbone=Gemma-3-4B2026.01 | 0.323 | 0.102 | — | — | |
| R_BayesBackbone=Qwen3-4B2026.01 | 0.32 | 0.214 | — | — | |
| R_BayesBackbone=Gemma-3-12B2026.01 | 0.313 | 0.21 | — | — | |
| R_BayesBackbone=Qwen3-30B-A3B2026.01 | 0.311 | 0.178 | — | — | |
| COCOABackbone=Gemma-3-4B2026.01 | 0.292 | 0.153 | — | — | |
| COCOABackbone=Qwen3-4B2026.01 | 0.29 | 0.133 | — | — | |
| COCOABackbone=Qwen3-30B-A3B2026.01 | 0.289 | 0.139 | — | — | |
| KLEBackbone=Qwen3-30B-A3B2026.01 | 0.28 | 0.065 | — | — | |
| SARBackbone=Gemma-3-12B2026.01 | 0.267 | 0.141 | — | — | |
| COCOABackbone=Gemma-3-12B2026.01 | 0.259 | 0.093 | — | — | |
| KLEBackbone=Qwen3-4B2026.01 | 0.257 | 0.068 | — | — | |
| SARBackbone=Qwen3-30B-A3B2026.01 | 0.25 | 0.114 | — | — | |
| SARBackbone=Qwen3-4B2026.01 | 0.246 | 0.115 | — | — | |
| KLEBackbone=Gemma-3-12B2026.01 | 0.228 | 0.047 | — | — | |
| P(True)Backbone=Qwen3-4B2026.01 | 0.226 | 0.208 | — | — | |
| MSPBackbone=Qwen3-4B2026.01 | 0.218 | 0.131 | — | — | |
| MSPBackbone=Qwen3-30B-A3B2026.01 | 0.194 | 0.114 | — | — | |
| MSPBackbone=Gemma-3-12B2026.01 | 0.178 | 0.061 | — | — | |
| MSPBackbone=Gemma-3-4B2026.01 | 0.168 | 0.081 | — | — | |
| SEBackbone=Gemma-3-4B2026.01 | 0.15 | 0.048 | — | — | |
| P(True)Backbone=Gemma-3-12B2026.01 | 0.147 | -0.04 | — | — | |
| SEBackbone=Qwen3-4B2026.01 | 0.144 | 0.037 | — | — | |
| SEBackbone=Gemma-3-12B2026.01 | 0.08 | 0.017 | — | — | |
| P(True)Backbone=Gemma-3-4B2026.01 | 0.075 | -0.044 | — | — | |
| SEBackbone=Qwen3-30B-A3B2026.01 | 0.059 | 0.035 | — | — | |
| P(True)Backbone=Qwen3-30B-A3B2026.01 | 0.032 | -0.017 | — | — | |
| COCOAModel=Gemma-3-4B2026.01 | — | — | 0.59 | 0.536 | |
| COCOAModel=Gemma-3-12B2026.01 | — | — | 0.585 | 0.526 | |
| COCOAModel=Qwen3-4B2026.01 | — | — | 0.594 | 0.537 | |
| COCOAModel=Qwen3-30B-A3B2026.01 | — | — | 0.598 | 0.541 | |
| KLEModel=Gemma-3-4B2026.01 | — | — | 0.609 | 0.541 | |
| KLEModel=Gemma-3-12B2026.01 | — | — | 0.588 | 0.524 | |
| KLEModel=Qwen3-4B2026.01 | — | — | 0.596 | 0.523 | |
| KLEModel=Qwen3-30B-A3B2026.01 | — | — | 0.6 | 0.521 | |
| MSPModel=Gemma-3-4B2026.01 | — | — | 0.565 | 0.521 | |
| MSPModel=Gemma-3-12B2026.01 | — | — | 0.571 | 0.519 | |
| MSPModel=Qwen3-4B2026.01 | — | — | 0.576 | 0.536 | |
| MSPModel=Qwen3-30B-A3B2026.01 | — | — | 0.579 | 0.535 | |
| P(True)Model=Gemma-3-4B2026.01 | — | — | 0.537 | 0.484 | |
| P(True)Model=Gemma-3-12B2026.01 | — | — | 0.557 | 0.494 | |
| P(True)Model=Qwen3-4B2026.01 | — | — | 0.525 | 0.55 | |
| P(True)Model=Qwen3-30B-A3B2026.01 | — | — | 0.518 | 0.503 | |
| RBayesModel=Gemma-3-4B2026.01 | — | — | 0.608 | 0.565 | |
| RBayesModel=Gemma-3-12B2026.01 | — | — | 0.595 | 0.562 | |
| RBayesModel=Qwen3-4B2026.01 | — | — | 0.605 | 0.558 | |
| RBayesModel=Qwen3-30B-A3B2026.01 | — | — | 0.599 | 0.548 | |
| SARModel=Gemma-3-4B2026.01 | — | — | 0.608 | 0.554 | |
| SARModel=Gemma-3-12B2026.01 | — | — | 0.586 | 0.544 | |
| SARModel=Qwen3-4B2026.01 | — | — | 0.582 | 0.531 | |
| SARModel=Qwen3-30B-A3B2026.01 | — | — | 0.584 | 0.53 | |
| SEModel=Gemma-3-4B2026.01 | — | — | 0.543 | 0.524 | |
| SEModel=Gemma-3-12B2026.01 | — | — | 0.519 | 0.508 | |
| SEModel=Qwen3-4B2026.01 | — | — | 0.535 | 0.512 | |
| SEModel=Qwen3-30B-A3B2026.01 | — | — | 0.517 | 0.508 |