Prompt Optimization on Causal Judgment
64.4AccuracyTRAS
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| TRASLLM=GPT-3.5-turbo, Seeds=5, Iterations=15, Warmup steps (τwarmup)=3, K=62025.07 | 64.4 | 0.008 | 2.162 | |
| Baseline + TRLLM=GPT-3.5-turbo, Seeds=5, Iterations=15, Warmup steps (τwarmup)=32025.07 | 63.6 | 0.04 | 1.336 | |
| Baseline + MCSALLM=GPT-3.5-turbo, Seeds=5, Iterations=15, Warmup steps (τwarmup)=3, K=62025.07 | 60.8 | 0.02 | 1.687 | |
| BaselineLLM=GPT-3.5-turbo, Seeds=5, Iterations=152025.07 | 58.6 | — | — |