Attribute-controlled Machine Translation on MT-GENEVAL (aggregated)
0.396BLEUAdapted MT
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Adapted MTPrompting Setting=Same-Language2023.05 | 0.396 | 0.75 | 0.842 | 0.864 | |
| BLOOM 175B RAMPBackbone=BLOOM 175B, Prompting Setting=Same-Language, Variant=RAMP, In-context Examples=162023.05 | 0.343 | 0.699 | 0.817 | 0.818 | |
| BLOOM 175B baseBackbone=BLOOM 175B, Prompting Setting=Same-Language, Variant=base, In-context Examples=162023.05 | 0.333 | 0.679 | 0.748 | 0.704 | |
| BLOOM 175B +markBackbone=BLOOM 175B, Prompting Setting=Same-Language, Variant=attribute marking (+mark), In-context Examples=162023.05 | 0.331 | 0.674 | 0.759 | 0.725 | |
| BLOOM 175B RAMP (Cross-Lingual)Backbone=BLOOM 175B, Prompting Setting=Cross-Lingual, Variant=RAMP, In-context Examples=162023.05 | 0.294 | 0.502 | 0.788 | 0.673 | |
| BLOOM 175B base (Cross-Lingual)Backbone=BLOOM 175B, Prompting Setting=Cross-Lingual, Variant=base, In-context Examples=162023.05 | 0.285 | 0.469 | 0.777 | 0.633 | |
| XGLM 7.5B RAMPBackbone=XGLM 7.5B, Prompting Setting=Same-Language, Variant=RAMP, In-context Examples=162023.05 | 0.248 | 0.473 | 0.836 | 0.82 | |
| XGLM 7.5B baseBackbone=XGLM 7.5B, Prompting Setting=Same-Language, Variant=base, In-context Examples=162023.05 | 0.237 | 0.445 | 0.79 | 0.727 | |
| XGLM 7.5B +markBackbone=XGLM 7.5B, Prompting Setting=Same-Language, Variant=attribute marking (+mark), In-context Examples=162023.05 | 0.237 | 0.444 | 0.789 | 0.732 |