Question Answering on TriviaQA (Cohen's d)
0.7498Cohen's dDebiased-Plausibility
Evaluation Results
| Method | Links | |
|---|---|---|
| Debiased-PlausibilityCategory=Entropy-Plausibility, alpha=0.88, N=62026.05 | 0.7498 | |
| PlausibilityCategory=Entropy-Plausibility, N=42026.05 | 0.6511 | |
| Debiased-PlausibilityCategory=Avg-Plausibility, alpha=0.57, N=162026.05 | 0.564 | |
| PlausibilityCategory=Avg-Plausibility, N=162026.05 | 0.5061 | |
| LLaMA 3.3 70bCategory=Uncertainty-based2026.05 | 0.4823 | |
| LLaMA 3.3 70bCategory=Prompt-based2026.05 | 0.4566 | |
| Retrieval ComplexityCategory=Retriever-based2026.05 | 0.4394 | |
| LLaMA 3.1 8bCategory=Prompt-based2026.05 | 0.361 | |
| LLaMA 3.1 8bCategory=Uncertainty-based2026.05 | 0.2211 | |
| Gunning-FogCategory=Readability2026.05 | -0.0963 | |
| Flesch-KincaidCategory=Readability2026.05 | -0.2689 | |
| PopQACategory=Popularity, N=162026.05 | -0.2702 |