| ############################################## | |
| step: final | |
| Table for all datasets except crossNER | |
| ACE 2004 : 36.7% | |
| ACE 2005 : 37.1% | |
| AnatEM : 27.6% | |
| Broad Tweet Corpus : 67.1% | |
| CoNLL 2003 : 71.3% | |
| FabNER : 33.3% | |
| FindVehicle : 39.3% | |
| GENIA_NER : 55.4% | |
| HarveyNER : 24.5% | |
| MultiNERD : 73.4% | |
| Ontonotes : 46.4% | |
| PolyglotNER : 51.9% | |
| TweetNER7 : 50.2% | |
| WikiANN en : 67.8% | |
| WikiNeural : 79.4% | |
| bc2gm : 41.9% | |
| bc4chemd : 66.6% | |
| bc5cdr : 72.6% | |
| ncbi : 71.0% | |
| Average : 53.3% | |
| Table for zero-shot benchmark | |
| CrossNER_AI : 61.5% | |
| CrossNER_literature : 69.1% | |
| CrossNER_music : 76.7% | |
| CrossNER_politics : 83.3% | |
| CrossNER_science : 68.7% | |
| mit-movie : 55.2% | |
| mit-restaurant : 50.6% | |
| Average : 66.5% | |
| ############################################## | |