impresso-project/ner-hipe2020-mmbert-base-final

Impresso HIPE-2022 NER model

Token-classification (NER) model fine-tuned from jhu-clsp/mmBERT-base on the impresso-project/ner-augmentation dataset (HIPE-2022, NE-COARSE-LIT, IOB2). Part of the Impresso NER pipeline.

• Languages: fr, de, en
• Base model: jhu-clsp/mmBERT-base
• Training data: impresso-project/ner-augmentation
• Label scheme: IOB2 over NE-COARSE-LIT (pers / org / loc / prod / time).

Test-set results (seqeval, entity-level)

metric	value
hipe2020/de_f1	0.7950
hipe2020/de_fuzzy_f1	0.8676
hipe2020/de_fuzzy_loc_f1	0.9238
hipe2020/de_fuzzy_org_f1	0.6129
hipe2020/de_fuzzy_pers_f1	0.8851
hipe2020/de_fuzzy_prod_f1	0.7048
hipe2020/de_fuzzy_time_f1	0.9474
hipe2020/de_loc_f1	0.8730
hipe2020/de_org_f1	0.5690
hipe2020/de_pers_f1	0.7601
hipe2020/de_prod_f1	0.6078
hipe2020/de_time_f1	0.8085
hipe2020/en_f1	0.6949
hipe2020/en_fuzzy_f1	0.7944
hipe2020/en_fuzzy_loc_f1	0.8280
hipe2020/en_fuzzy_org_f1	0.5769
hipe2020/en_fuzzy_pers_f1	0.8769
hipe2020/en_fuzzy_prod_f1	0.5882
hipe2020/en_fuzzy_time_f1	0.9412
hipe2020/en_loc_f1	0.7692
hipe2020/en_org_f1	0.3862
hipe2020/en_pers_f1	0.7508
hipe2020/en_prod_f1	0.6000
hipe2020/en_time_f1	0.7647
hipe2020/fr_f1	0.8254
hipe2020/fr_fuzzy_f1	0.9018
hipe2020/fr_fuzzy_loc_f1	0.9329
hipe2020/fr_fuzzy_org_f1	0.6891
hipe2020/fr_fuzzy_pers_f1	0.9238
hipe2020/fr_fuzzy_prod_f1	0.8288
hipe2020/fr_fuzzy_time_f1	0.8870
hipe2020/fr_loc_f1	0.8773
hipe2020/fr_org_f1	0.6328
hipe2020/fr_pers_f1	0.8213
hipe2020/fr_prod_f1	0.7358
hipe2020/fr_time_f1	0.5926
hipe2020_f1	0.7964
hipe2020_fuzzy_f1	0.8743
hipe2020_fuzzy_loc_f1	0.9177
hipe2020_fuzzy_org_f1	0.6349
hipe2020_fuzzy_pers_f1	0.9037
hipe2020_fuzzy_prod_f1	0.7440
hipe2020_fuzzy_time_f1	0.9180
hipe2020_loc_f1	0.8636
hipe2020_org_f1	0.5529
hipe2020_pers_f1	0.7906
hipe2020_precision	0.7999
hipe2020_prod_f1	0.6639
hipe2020_recall	0.7929
hipe2020_time_f1	0.7034

License

Inherits the CC BY-NC-SA 4.0 license of the underlying HIPE-2022 data.