Literary Translation on par3 dev evaluation (full)
0.773COMETKiwiGemini
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| GeminiPipeline=P2, Human eval.=false2026.06 | 0.773 | 0.805 | 5.182 | 4.758 | |
| GeminiPipeline=P1, Human eval.=false2026.06 | 0.772 | 0.804 | 5.243 | 4.813 | |
| GPT-5.4Pipeline=P2, Human eval.=false2026.06 | 0.768 | 0.8 | 5.482 | 5.02 | |
| GPT-5.4Pipeline=P1, Human eval.=false2026.06 | 0.766 | 0.798 | 5.542 | 5.073 | |
| AgentsPipeline=P3, Human eval.=true2026.06 | 0.766 | 0.793 | 5.656 | 5.109 | |
| HTHuman eval.=true2026.06 | 0.726 | — | — | 5.726 |