Multilabel Classification on CLIP (test)
80.9Micro F1TTP-BERT+Context (2M)
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| TTP-BERT+Context (2M)Pre-training data size=2M sentences, Use Context=true2021.06 | 80.9 | 95.7 | 66 | 97.3 | 86.5 | |
| TTP-BERT+Context (250k)Pre-training data size=250k sentences, Use Context=true2021.06 | 80.7 | 96.2 | 66.8 | 97.5 | 86.6 | |
| TTP-BERT+Context (500k)Pre-training data size=500k sentences, Use Context=true2021.06 | 80.3 | 95.3 | 67.1 | 97.6 | 86.2 | |
| TTP-BERT+Context (1M)Pre-training data size=1M sentences, Use Context=true2021.06 | 80.2 | 95.9 | 65.4 | 97.4 | 85.7 | |
| MIMIC-DNote-BERT+ContextUse Context=true2021.06 | 79.6 | 95.8 | 66.1 | 97.7 | 85.6 | |
| MIMIC-Full-BERT+ContextUse Context=true2021.06 | 79.4 | 95.4 | 64.1 | 97.2 | 85.7 | |
| BERT+ContextUse Context=true2021.06 | 79.1 | 94.7 | 63.5 | 97.1 | 85.6 | |
| MIMIC-DNote-BERTUse Context=false2021.06 | 76.7 | 97.2 | 63.1 | 96.7 | 83.4 | |
| MIMIC-Full-BERTUse Context=false2021.06 | 76.5 | 97.1 | 62.4 | 96.6 | 83.2 | |
| BERTUse Context=false2021.06 | 75.8 | 96.2 | 59.3 | 96.3 | 82.7 | |
| CNNVectors=BioWordVec, Use Context=false2021.06 | 72.3 | 96.4 | 54 | 96.2 | 81 | |
| Bag-of-words+TFIDFRegularization=L1, Use Context=false2021.06 | 70.9 | 95.8 | 51.2 | 93.7 | 78.3 |