Span-level Hallucination Detection on RagTruth-Avg (test)
76.63F1 ScoreFourier-high
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Fourier-highModel=Mistral-7B, sliding-window=82026.02 | 76.63 | 88.12 | |
| Wavelet-HighModel=Mistral-7B, sliding-window=82026.02 | 75.97 | 87.42 | |
| Lookback-lensModel=Mistral-7B, sliding-window=82026.02 | 75.54 | 87.64 | |
| LaplacianModel=Mistral-7B, sliding-window=82026.02 | 73.22 | 84.95 | |
| Fourier-highModel=LLaMA-13B, sliding-window=82026.02 | 70.63 | 85.85 | |
| Fourier-highModel=LLaMA-7B, sliding-window=82026.02 | 70.03 | 84.12 | |
| Wavelet-HighModel=LLaMA-13B, sliding-window=82026.02 | 69.53 | 84.48 | |
| Wavelet-HighModel=LLaMA-7B, sliding-window=82026.02 | 69.11 | 81.17 | |
| LaplacianModel=LLaMA-13B, sliding-window=82026.02 | 68.19 | 80.39 | |
| Lookback-lensModel=LLaMA-13B, sliding-window=82026.02 | 67.81 | 81.83 | |
| Lookback-lensModel=LLaMA-7B, sliding-window=82026.02 | 67.73 | 78.84 | |
| LaplacianModel=LLaMA-7B, sliding-window=82026.02 | 67.47 | 78.77 |