Prompt optimization on CB
85.7AccuracyTRAS
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| TRASLLM=GPT-3.5-turbo, Seeds=5, Iterations=15, Warmup steps (τwarmup)=4, K=62025.07 | 85.7 | 0.008 | 2.495 | |
| Baseline + MCSALLM=GPT-3.5-turbo, Seeds=5, Iterations=15, Warmup steps (τwarmup)=4, K=62025.07 | 84.2 | 0.032 | 1.61 | |
| Baseline + TRLLM=GPT-3.5-turbo, Seeds=5, Iterations=15, Warmup steps (τwarmup)=42025.07 | 84.2 | 0.049 | 1.402 | |
| BaselineLLM=GPT-3.5-turbo, Seeds=5, Iterations=152025.07 | 81.7 | — | — |