This is the code for An Embarrassingly Easy but Strong Baseline for Nested Named Entity Recognition
Using the ConLL 2003 dataset with bert-base-cased pretrained embeddings and a random seed of 43, the best test set F1 score is 91.88
| seed | P. | R. | F1 |
|---|---|---|---|
| 13 | 92.35 | 91.45 | 91.90 |
| 43(1) | 91.71 | 92.05 | 91.88 |
| 43(2) | 91.87 | 91.84 | 91.85 |