Chinese Language Generation on Chinese Dataset
4.6Usefulness ScoreBucket-SFT
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Bucket-SFTn=202026.06 | 4.6 | 4.95 | 5 | 4.85 | |
| Full-SFTn=202026.06 | 4.45 | 4.65 | 4.55 | 4.55 | |
| BaseLMn=192026.06 | 3.47 | 3.89 | 3.84 | 3.74 | |
| DPOn=202026.06 | 3.05 | 3.6 | 2.95 | 3.2 |