Prompt Optimization on DQA
82.3Best ScoreSCULPT
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| SCULPTModel=GPT-5, Optimization Rounds (R)=82026.06 | 82.3 | 81.1 | 1,240 | 6.6 | 78 | 52 | 2 | |
| CRAFTModel=GPT-5, Optimization Rounds (R)=82026.06 | 77.7 | 75.1 | 225 | 34.5 | 77 | 20 | 3 | |
| InitialModel=GPT-5, Optimization Rounds (R)=82026.06 | 76.6 | 76.6 | 512 | 15 | — | — | 1 | |
| DISTILLModel=GPT-5, Optimization Rounds (R)=82026.06 | 76.6 | 74.6 | 224 | 34.2 | 73 | 41 | 2 | |
| WPRO0.5Model=GPT-5, Optimization Rounds (R)=82026.06 | 76 | 73.6 | 251 | 30.3 | 72 | 47 | 3 | |
| LLMLinguaModel=GPT-5, Optimization Rounds (R)=82026.06 | 76 | 74.6 | 298 | 25.5 | 75 | 40 | 3 | |
| WPRO0.3Model=GPT-5, Optimization Rounds (R)=82026.06 | 73.7 | 70.5 | 190 | 38.8 | 74 | 41 | 2 | |
| WPRO0.7Model=GPT-5, Optimization Rounds (R)=82026.06 | 70.3 | 68.5 | 228 | 30.8 | 70 | 65 | 2 |