Few-NERD: A Few-Shot Named Entity Recognition Dataset

Benchmark Model Rank Results
few-shot-ner-on-few-nerd-interStructShot#105 way 1~2 shot: 51.88±0.695 way 5~10 shot: 57.32±0.63
few-shot-ner-on-few-nerd-interNNShot#115 way 1~2 shot: 47.24±1.005 way 5~10 shot: 55.64±0.63
few-shot-ner-on-few-nerd-interProtoBERT#125 way 1~2 shot: 38.83±1.495 way 5~10 shot: 58.79±0.44
few-shot-ner-on-few-nerd-intraStructShot#105 way 1~2 shot: 30.21±0.905 way 5~10 shot: 38.00±1.29
few-shot-ner-on-few-nerd-intraNNShot#115 way 1~2 shot: 25.78±0.915 way 5~10 shot: 36.18±0.79
few-shot-ner-on-few-nerd-intraProtoBERT#125 way 1~2 shot: 20.76±0.845 way 5~10 shot: 42.54±0.94
named-entity-recognition-on-few-nerd-supBERT-Tagger#6F1-Measure: 67.13Precision: 65.56Recall: 68.78