| ############################################## | |
| step: final | |
| Table for all datasets except crossNER | |
| ACE 2004 : 38.5% | |
| ACE 2005 : 39.6% | |
| AnatEM : 48.8% | |
| Broad Tweet Corpus : 64.5% | |
| CoNLL 2003 : 66.0% | |
| FabNER : 35.8% | |
| FindVehicle : 48.8% | |
| GENIA_NER : 59.4% | |
| HarveyNER : 28.0% | |
| MultiNERD : 61.8% | |
| Ontonotes : 39.7% | |
| PolyglotNER : 48.5% | |
| TweetNER7 : 52.3% | |
| WikiANN en : 69.6% | |
| WikiNeural : 75.0% | |
| bc2gm : 64.6% | |
| bc4chemd : 64.6% | |
| bc5cdr : 74.1% | |
| ncbi : 74.9% | |
| Average : 55.5% | |
| Table for zero-shot benchmark | |
| CrossNER_AI : 59.1% | |
| CrossNER_literature : 72.4% | |
| CrossNER_music : 76.0% | |
| CrossNER_politics : 83.1% | |
| CrossNER_science : 66.6% | |
| mit-movie : 65.2% | |
| mit-restaurant : 53.6% | |
| Average : 68.0% | |
| ############################################## | |