Hallucination Regression on PHOENIX 2014T
-0.189PearsonPerplexity
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| PerplexityModel Type=Gloss-based (GB)2025.10 | -0.189 | -0.229 | 0.296 | |
| Token entropyModel Type=Gloss-based (GB)2025.10 | -0.275 | -0.273 | 0.342 | |
| ConfidenceModel Type=Gloss-based (GB)2025.10 | -0.344 | -0.308 | 0.43 | |
| PerplexityModel Type=Gloss-free (GF)2025.10 | -0.4 | -0.542 | 0.609 | |
| GroundingModel Type=Gloss-based (GB)2025.10 | -0.458 | -0.4 | 0.493 | |
| Token entropyModel Type=Gloss-free (GF)2025.10 | -0.468 | -0.544 | 0.625 | |
| METAModel Type=Gloss-based (GB)2025.10 | -0.61 | -0.574 | 0.637 | |
| ConfidenceModel Type=Gloss-free (GF)2025.10 | -0.612 | -0.578 | 0.637 | |
| GroundingModel Type=Gloss-free (GF)2025.10 | -0.623 | -0.59 | 0.65 | |
| METAModel Type=Gloss-free (GF)2025.10 | -0.65 | -0.613 | 0.675 |