Medical Text Classification on MTSamples held-out (test)
27.9Macro F1 ScoreCLiGNet
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| CLiGNetcalibration=none2026.03 | 27.9 | 36.2 | 36.2 | 0.032 | — | |
| Longformer + BRmulti_label_strategy=Binary Relevance2026.03 | 26.2 | 33.6 | 33.6 | 0.033 | — | |
| CLiGNetcalibration=full (Platt scaling)2026.03 | 24 | 33.8 | 33.8 | 0.04 | 0.007 | |
| ClinBERT + OvRmulti_label_strategy=One-vs-Rest2026.03 | 21.1 | 34.4 | 34.4 | 0.033 | — | |
| TF-IDF + LRclassifier=Logistic Regression2026.03 | 17.3 | 10.5 | 10.5 | 0.045 | — | |
| BioBERT + OvRmulti_label_strategy=One-vs-Rest2026.03 | 17.2 | 31.8 | 31.8 | 0.034 | — | |
| TF-IDF + SVCclassifier=Support Vector Classifier2026.03 | 9.9 | 15.2 | 15.2 | 0.042 | — |