Hallucination Detection on TruthfulQA (Specific AUROC Variants)
0.61AUROC (phi C75)Transport Diagnostics
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Transport DiagnosticsModel=GPT-2, Evaluation=length-controlled2026.05 | 0.61 | 0.57 | 0.6 | — | — | — | — | |
| Transport DiagnosticsModel=Pythia-160M, Evaluation=length-controlled2026.05 | 0.57 | 0.58 | 0.57 | — | — | — | — | |
| Transport DiagnosticsModel=BERT, Evaluation=length-controlled2026.05 | 0.55 | 0.56 | 0.55 | — | — | — | — | |
| Transport DiagnosticsModel=Flan-T5 (dec), Evaluation=length-controlled2026.05 | 0.54 | 0.55 | 0.53 | — | — | — | — | |
| Transport DiagnosticsModel=Flan-T5 (cross), Evaluation=length-controlled2026.05 | 0.53 | 0.58 | 0.57 | — | — | — | — | |
| EigenTrackModel=GPT-2, Evaluation=length-controlled2026.05 | — | — | — | — | — | — | 0.5 | |
| EigenTrackModel=BERT, Evaluation=length-controlled2026.05 | — | — | — | — | — | — | 0.52 | |
| EigenTrackModel=Pythia-160M, Evaluation=length-controlled2026.05 | — | — | — | — | — | — | 0.52 | |
| EigenTrackModel=Flan-T5 (cross), Evaluation=length-controlled2026.05 | — | — | — | — | — | — | 0.53 | |
| EigenTrackModel=Flan-T5 (dec), Evaluation=length-controlled2026.05 | — | — | — | — | — | — | 0.53 | |
| LLM-CheckModel=GPT-2, Evaluation=length-controlled2026.05 | — | — | — | 0.55 | 0.63 | 0.58 | — | |
| LLM-CheckModel=BERT, Evaluation=length-controlled2026.05 | — | — | — | 0.57 | 0.62 | — | — | |
| LLM-CheckModel=Pythia-160M, Evaluation=length-controlled2026.05 | — | — | — | 0.52 | 0.63 | 0.52 | — | |
| LLM-CheckModel=Flan-T5 (cross), Evaluation=length-controlled2026.05 | — | — | — | 0.59 | — | — | — | |
| LLM-CheckModel=Flan-T5 (dec), Evaluation=length-controlled2026.05 | — | — | — | 0.6 | 0.59 | 0.56 | — |