Question Generation on RACE (test)
1.95FluencyQG_sft
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| QG_sftMethod=Supervised Fine-Tuned2025.10 | 1.95 | 99 | 72 | |
| RACE DatasetMethod=Human-created questions (Reference)2025.10 | 1.94 | 98 | 91 | |
| QG_dpoMethod=Direct Preference Optimized2025.10 | 1.94 | 97 | 76 |