Form Success Rate (FSR) evaluation on Llama-2 variants ProFLingo evaluation suite
100Form Success Rate (Llama-2)ProFLingo
Evaluation Results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| ProFLingoTask Combination=Base2025.05 | 100 | 24 | 52 | 42 | 28 | 20 | 14 | 28 | 100 | 100 | |
| ProFLingo+Homologous ModelTask Combination=Homologous Model2025.05 | 100 | 32 | 68 | 54 | 42 | 28 | 18 | 42 | 100 | 100 | |
| ProFLingo+Chat TemplateTask Combination=Chat Template2025.05 | 100 | 30 | 70 | 52 | 38 | 26 | 16 | 48 | 100 | 100 | |
| ProFLingo+S2.7-P+[D+Z]Task Combination=S2.7-P+[D+Z]2025.05 | 100 | 48 | 84 | 76 | 72 | 44 | 28 | 82 | 100 | 100 |