Hallucination Detection on PubMedQA (test)
75MF1 ScoreCALAMRFLOW
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| CALAMRFLOWAveraged over=5 random seeds2026.05 | 75 | 61.5 | 84.1 | 71 | 75.6 | 77.5 | |
| CALAMRFLOWseeds=5, test-set size (N)=6202026.05 | 75 | — | — | 71 | 75.6 | — | |
| Human (single)2026.05 | 72.2 | — | — | — | 78 | — | |
| Human (single)test-set size (N)=6202026.05 | 72.2 | — | — | — | 78 | — | |
| ESIM w/ BioELMo2026.05 | 31.1 | — | — | — | 54 | — | |
| ESIM w/ BioELMotest-set size (N)=6202026.05 | 31.1 | — | — | — | 54 | — | |
| BioBERT2026.05 | 28.7 | — | — | — | 57.3 | — | |
| BioBERTtest-set size (N)=6202026.05 | 28.7 | — | — | — | 57.3 | — | |
| BiLSTM2026.05 | 23.9 | — | — | — | 55.2 | — | |
| BiLSTMtest-set size (N)=6202026.05 | 23.9 | — | — | — | 55.2 | — |