Masked Language Modeling on Turkish Datasets (Composite) (test)
69.57MLM Avg (%)boun-tabilab/TabiBERT
Evaluation results
| Method | Links | |
|---|---|---|
| boun-tabilab/TabiBERTMasking Strategy=80-10-10, Evaluation Corpus Case Sensitivity=mixed-case2026.01 | 69.57 | |
| newmindai/Mursit-LargeMasking Strategy=80-10-10, Evaluation Corpus Case Sensitivity=mixed-case2026.01 | 67.25 | |
| ytu-ce-cosmos/turkish-large-bert-casedMasking Strategy=80-10-10, Evaluation Corpus Case Sensitivity=mixed-case2026.01 | 65.03 | |
| dbmdz/bert-base-turkish-casedMasking Strategy=80-10-10, Evaluation Corpus Case Sensitivity=mixed-case2026.01 | 64.98 | |
| newmindai/Mursit-BaseMasking Strategy=80-10-10, Evaluation Corpus Case Sensitivity=mixed-case2026.01 | 64.05 | |
| KocLab-Bilkent/BERTurk-LegalMasking Strategy=80-10-10, Evaluation Corpus Case Sensitivity=mixed-case2026.01 | 54.1 | |
| ytu-ce-cosmos/turkish-base-bert-uncasedMasking Strategy=80-10-10, Evaluation Corpus Case Sensitivity=uncased2026.01 | 52.69 |