metadata
title: README
emoji: 📚
colorFrom: indigo
colorTo: purple
sdk: static
pinned: false
hmBERT
Historical Multilingual Language Models for Named Entity Recognition. The following languages are covered by hmBERT:
- English (British Library Corpus - Books)
- German (Europeana Newspaper)
- French (Europeana Newspaper)
- Finnish (Europeana Newspaper)
- Swedish (Europeana Newspaper)
More details can be found in our GitHub repository and in our hmBERT paper.
Leaderboard
We test our pretrained language models on various datasets from HIPE-2020, HIPE-2022 and Europeana. The following table shows an overview of used datasets:
| Language | Datasets |
|---|---|
| English | AjMC - TopRes19th |
| German | AjMC - NewsEye - HIPE-2020 |
| French | AjMC - ICDAR-Europeana - LeTemps - NewsEye - HIPE-2020 |
| Finnish | NewsEye |
| Swedish | NewsEye |
| Dutch | ICDAR-Europeana |
Results:
| Model | English AjMC | German AjMC | French AjMC | German NewsEye | French NewsEye | Finnish NewsEye | Swedish NewsEye | Dutch ICDAR | French ICDAR | French LeTemps | English TopRes19th | German HIPE-2020 | French HIPE-2020 | Avg. |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| hmBERT (32k) Schweter et al. | 85.36 ± 0.94 | 89.08 ± 0.09 | 85.10 ± 0.60 | 39.65 ± 1.01 | 81.47 ± 0.36 | 77.28 ± 0.37 | 82.85 ± 0.83 | 82.11 ± 0.61 | 77.21 ± 0.16 | 65.73 ± 0.56 | 80.94 ± 0.86 | 79.18 ± 0.38 | 83.47 ± 0.80 | 77.65 |
| hmTEAMS | 86.41 ± 0.36 | 88.64 ± 0.42 | 85.41 ± 0.67 | 41.51 ± 2.82 | 83.20 ± 0.79 | 79.27 ± 1.88 | 82.78 ± 0.60 | 88.21 ± 0.39 | 78.03 ± 0.39 | 66.71 ± 0.46 | 81.36 ± 0.59 | 80.15 ± 0.60 | 86.07 ± 0.49 | 79.06 |
Acknowledgements
We thank Luisa März, Katharina Schmid and Erion Çano for their fruitful discussions about Historical Language Models.
Research supported with Cloud TPUs from Google's TPU Research Cloud (TRC). Many Thanks for providing access to the TPUs ❤️