Named Entity Recognition on JNLPBA (test)
83.8Macro F1 (span-level)Unified-00
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Unified-00Model Type=Proposed Unified Model2019.09 | 83.8 | — | — | |
| Spark NLP2021.04 | 81.29 | — | — | |
| Unified-01Model Type=Proposed Unified Model2019.09 | 80.9 | — | — | |
| Unified-11Model Type=Proposed Unified Model2019.09 | 80.9 | — | — | |
| BioBERT-v1.1Number of Parameters=> 60M2022.09 | 80.33 | 74.81 | 86.72 | |
| DistilBioBERTNumber of Parameters=> 60M2022.09 | 79.97 | 74.36 | 86.49 | |
| CompactBioBERTNumber of Parameters=> 60M2022.09 | 79.88 | 73.84 | 86.98 | |
| DistilBERTNumber of Parameters=> 60M2022.09 | 79.14 | 73.37 | 85.9 | |
| TinyBioBERTNumber of Parameters=15M2022.09 | 78.75 | 72.58 | 86.07 | |
| SOTA (Collier and Kim, 2004)2019.03 | 78.58 | — | — | |
| BioBERTtraining_mode=finetune2019.03 | 77.59 | — | — | |
| BioBERT2021.04 | 77.59 | 72.24 | 83.26 | |
| SciBERTtraining_mode=finetune2019.03 | 77.28 | — | — | |
| SCIBERTEvaluation Protocol=Finetune2019.03 | 77.28 | — | — | |
| BioFLAIR2021.04 | 77.03 | — | — | |
| BERT-BaseEvaluation Protocol=Finetune2019.03 | 76.09 | — | — | |
| SCIBERTEvaluation Protocol=Frozen2019.03 | 75.77 | — | — | |
| MLM baselineCtx=8192, Model Size=Large 350M, Training Tokens=50B tokens2026.05 | 75.5 | — | — | |
| CLM detourCtx=8192, Model Size=Large 350M, Training Tokens=50B tokens2026.05 | 75.3 | — | — | |
| PubMedBERTCtx=512, Model Size=Base2026.05 | 74.9 | — | — | |
| MLM baselineCtx=8192, Model Size=Base 150M, Training Tokens=10B tokens2026.05 | 74.9 | — | — | |
| MLM baselineCtx=8192, Model Size=Base 150M, Training Tokens=50B tokens2026.05 | 74.9 | — | — | |
| BioClinical-ModernBERTCtx=8192, Model Size=Base2026.05 | 74.8 | — | — | |
| CLM detourCtx=8192, Model Size=Base 150M, Training Tokens=50B tokens2026.05 | 74.5 | — | — | |
| ModernBERT-baseCtx=8192, Model Size=Base2026.05 | 74.3 | — | — | |
| CLM detourCtx=8192, Model Size=Base 150M, Training Tokens=10B tokens2026.05 | 74.2 | — | — | |
| BERT-BaseEvaluation Protocol=Frozen2019.03 | 74.05 | — | — | |
| ELECTRAMedAveraging=5 runs with different seeds2021.04 | 73.65 | 69.33 | 78.56 | |
| MTM Wang et al. (2018)Model Type=Multi-task2019.09 | 73.5 | — | — | |
| MTM (ours)Model Type=Multi-task2019.09 | 73.5 | — | — | |
| STMModel Type=Single-task2019.09 | 72.7 | — | — | |
| MTM Crichton et al. (2017)Model Type=Multi-task2019.09 | 70.1 | — | — |