| ############################################## | |
| step: final | |
| Table for all datasets except crossNER | |
| ACE 2004 : 37.4% | |
| ACE 2005 : 38.1% | |
| AnatEM : 57.9% | |
| Broad Tweet Corpus : 63.6% | |
| CoNLL 2003 : 65.4% | |
| FabNER : 35.8% | |
| FindVehicle : 43.1% | |
| GENIA_NER : 51.6% | |
| HarveyNER : 24.5% | |
| MultiNERD : 66.9% | |
| Ontonotes : 47.7% | |
| PolyglotNER : 47.5% | |
| TweetNER7 : 51.2% | |
| WikiANN en : 70.5% | |
| WikiNeural : 71.7% | |
| bc2gm : 65.3% | |
| bc4chemd : 59.8% | |
| bc5cdr : 71.0% | |
| ncbi : 69.6% | |
| Average : 54.7% | |
| Table for zero-shot benchmark | |
| CrossNER_AI : 56.6% | |
| CrossNER_literature : 71.1% | |
| CrossNER_music : 72.2% | |
| CrossNER_politics : 81.6% | |
| CrossNER_science : 64.4% | |
| mit-movie : 62.3% | |
| mit-restaurant : 53.2% | |
| Average : 65.9% | |
| ############################################## |