Tabular Data Synthesis on Average of 4 Datasets Excluding Beijing
0.8225MLEGReaT-FT+
Evaluation results
| Method | Links | |
|---|---|---|
| GReaT-FT+Alignment Framework=Baseline2026.04 | 0.8225 | |
| TabNPO + KL penaltyAlignment Framework=NPO, KL Penalty=true2026.04 | 0.8228 | |
| TabNPO + Gradient diff.Alignment Framework=NPO, Gradient Difference=true2026.04 | 0.8241 | |
| TabKTO (base)Alignment Framework=KTO, Loss Function Variant=base2026.04 | 0.8242 | |
| TabDPO (base)Alignment Framework=DPO, Loss Function Variant=base2026.04 | 0.8246 | |
| TabDPO + KL penaltyAlignment Framework=DPO, KL Penalty=true2026.04 | 0.8249 | |
| TabNPO (base)Alignment Framework=NPO, Loss Function Variant=base2026.04 | 0.8257 | |
| TabDPO + Gradient diff.Alignment Framework=DPO, Gradient Difference=true2026.04 | 0.8266 | |
| GReaTAlignment Framework=Baseline2026.04 | 0.827 | |
| TabKTO + LogsigmoidAlignment Framework=KTO, Logsigmoid=true2026.04 | 0.8274 | |
| TabKTO + Logs. + Grad. diffAlignment Framework=KTO, Logsigmoid=true, Gradient Difference=true2026.04 | 0.8288 | |
| TabGRAA + Logs. + Grad. diffAlignment Framework=GRAA, Logsigmoid=true, Gradient Difference=true2026.04 | 0.8292 | |
| TabGRAA + LogsigmoidAlignment Framework=GRAA, Logsigmoid=true2026.04 | 0.8298 | |
| TabGRAA (base)Alignment Framework=GRAA, Loss Function Variant=base2026.04 | 0.8314 |