File size: 4,378 Bytes
1920e88 2443a7d 86d2ddf 1920e88 86d2ddf 2443a7d 86d2ddf |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 |
---
license: apache-2.0
language:
- en
- de
- fr
- fi
- sv
- nl
- nb
- nn
- 'no'
---
# hmTEAMS
[](https://github.com/stefan-it/hmTEAMS)
Historic Multilingual and Monolingual [TEAMS](https://aclanthology.org/2021.findings-acl.219/) Models.
The following languages are covered:
* English (British Library Corpus - Books)
* German (Europeana Newspaper)
* French (Europeana Newspaper)
* Finnish (Europeana Newspaper, Digilib)
* Swedish (Europeana Newspaper, Digilib)
* Dutch (Delpher Corpus)
* Norwegian (NCC Corpus)
# Architecture
We pretrain a "Training ELECTRA Augmented with Multi-word Selection"
([TEAMS](https://aclanthology.org/2021.findings-acl.219/)) model:
# Results
We perform experiments on various historic NER datasets, such as HIPE-2022 or ICDAR Europeana.
All details incl. hyper-parameters can be found [here](https://github.com/stefan-it/hmTEAMS/tree/main/bench).
## Small Benchmark
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 | Dataset | Additional Dataset |
|----------|--------------------------------------------------------------------------------------------------|----------------------------------------------------------------------------------|
| English | [AjMC](https://github.com/hipe-eval/HIPE-2022-data/blob/main/documentation/README-ajmc.md) | - |
| German | [AjMC](https://github.com/hipe-eval/HIPE-2022-data/blob/main/documentation/README-ajmc.md) | - |
| French | [AjMC](https://github.com/hipe-eval/HIPE-2022-data/blob/main/documentation/README-ajmc.md) | [ICDAR-Europeana](https://github.com/stefan-it/historic-domain-adaptation-icdar) |
| Finnish | [NewsEye](https://github.com/hipe-eval/HIPE-2022-data/blob/main/documentation/README-newseye.md) | - |
| Swedish | [NewsEye](https://github.com/hipe-eval/HIPE-2022-data/blob/main/documentation/README-newseye.md) | - |
| Dutch | [ICDAR-Europeana](https://github.com/stefan-it/historic-domain-adaptation-icdar) | - |
# Results
| Model | English AjMC | German AjMC | French AjMC | Finnish NewsEye | Swedish NewsEye | Dutch ICDAR | French ICDAR | Avg. |
|----------------------------------------------------------------------------------------|--------------|--------------|--------------|-----------------|-----------------|--------------|--------------|-----------|
| hmBERT (32k) [Schweter et al.](https://ceur-ws.org/Vol-3180/paper-87.pdf) | 85.36 ± 0.94 | 89.08 ± 0.09 | 85.10 ± 0.60 | 77.28 ± 0.37 | 82.85 ± 0.83 | 82.11 ± 0.61 | 77.21 ± 0.16 | 82.71 |
| hmTEAMS (Ours) | 86.41 ± 0.36 | 88.64 ± 0.42 | 85.41 ± 0.67 | 79.27 ± 1.88 | 82.78 ± 0.60 | 88.21 ± 0.39 | 78.03 ± 0.39 | **84.11** |
# Release
Our pretrained hmTEAMS model can be obtained from the Hugging Face Model Hub:
* [hmTEAMS Discriminator (**this model**)](https://huggingface.co/hmteams/teams-base-historic-multilingual-discriminator)
* [hmTEAMS Generator](https://huggingface.co/hmteams/teams-base-historic-multilingual-generator)
# Acknowledgements
We thank [Luisa März](https://github.com/LuisaMaerz), [Katharina Schmid](https://github.com/schmika) and
[Erion Çano](https://github.com/erionc) for their fruitful discussions about Historic Language Models.
Research supported with Cloud TPUs from Google's [TPU Research Cloud](https://sites.research.google/trc/about/) (TRC).
Many Thanks for providing access to the TPUs ❤️ |