---
language: de
license: mit
tags:
- flair
- token-classification
- sequence-tagger-model
base_model: hmbyt5-preliminary/byt5-small-historic-multilingual-span20-flax
widget:
- text: — Dramatiſch war der Stoff vor Sophokles von Äſchylos behandelt worden in
    den Θροῇσσαι , denen vielleicht in der Trilogie das Stüc>"OnJw» κοίσις vorherging
    , das Stück Σαλαμίνιαι folgte .
---

# Fine-tuned Flair Model on AjMC German NER Dataset (HIPE-2022)

This Flair model was fine-tuned on the
[AjMC German](https://github.com/hipe-eval/HIPE-2022-data/blob/main/documentation/README-ajmc.md)
NER Dataset using hmByT5 as backbone LM.

The AjMC dataset consists of NE-annotated historical commentaries in the field of Classics,
and was created in the context of the [Ajax MultiCommentary](https://mromanello.github.io/ajax-multi-commentary/)
project.

The following NEs were annotated: `pers`, `work`, `loc`, `object`, `date` and `scope`.

# Results

We performed a hyper-parameter search over the following parameters with 5 different seeds per configuration:

* Batch Sizes: `[8, 4]`
* Learning Rates: `[0.00015, 0.00016]`

And report micro F1-score on development set:

| Configuration     | Run 1        | Run 2        | Run 3        | Run 4        | Run 5        | Avg.         |
|-------------------|--------------|--------------|--------------|--------------|--------------|--------------|
| bs4-e10-lr0.00016 | [0.8892][1]  | [0.8913][2]  | [0.8867][3]  | [0.8843][4]  | [0.8828][5]  | 88.69 ± 0.31 |
| bs4-e10-lr0.00015 | [0.8786][6]  | [0.8793][7]  | [0.883][8]   | [0.8807][9]  | [0.8722][10] | 87.88 ± 0.36 |
| bs8-e10-lr0.00016 | [0.8602][11] | [0.8684][12] | [0.8643][13] | [0.8643][14] | [0.8623][15] | 86.39 ± 0.27 |
| bs8-e10-lr0.00015 | [0.8551][16] | [0.8707][17] | [0.8599][18] | [0.8609][19] | [0.8612][20] | 86.16 ± 0.51 |

[1]: https://hf.co/hmbench/hmbench-ajmc-de-hmbyt5-bs4-wsFalse-e10-lr0.00016-poolingfirst-layers-1-crfFalse-1
[2]: https://hf.co/hmbench/hmbench-ajmc-de-hmbyt5-bs4-wsFalse-e10-lr0.00016-poolingfirst-layers-1-crfFalse-2
[3]: https://hf.co/hmbench/hmbench-ajmc-de-hmbyt5-bs4-wsFalse-e10-lr0.00016-poolingfirst-layers-1-crfFalse-3
[4]: https://hf.co/hmbench/hmbench-ajmc-de-hmbyt5-bs4-wsFalse-e10-lr0.00016-poolingfirst-layers-1-crfFalse-4
[5]: https://hf.co/hmbench/hmbench-ajmc-de-hmbyt5-bs4-wsFalse-e10-lr0.00016-poolingfirst-layers-1-crfFalse-5
[6]: https://hf.co/hmbench/hmbench-ajmc-de-hmbyt5-bs4-wsFalse-e10-lr0.00015-poolingfirst-layers-1-crfFalse-1
[7]: https://hf.co/hmbench/hmbench-ajmc-de-hmbyt5-bs4-wsFalse-e10-lr0.00015-poolingfirst-layers-1-crfFalse-2
[8]: https://hf.co/hmbench/hmbench-ajmc-de-hmbyt5-bs4-wsFalse-e10-lr0.00015-poolingfirst-layers-1-crfFalse-3
[9]: https://hf.co/hmbench/hmbench-ajmc-de-hmbyt5-bs4-wsFalse-e10-lr0.00015-poolingfirst-layers-1-crfFalse-4
[10]: https://hf.co/hmbench/hmbench-ajmc-de-hmbyt5-bs4-wsFalse-e10-lr0.00015-poolingfirst-layers-1-crfFalse-5
[11]: https://hf.co/hmbench/hmbench-ajmc-de-hmbyt5-bs8-wsFalse-e10-lr0.00016-poolingfirst-layers-1-crfFalse-1
[12]: https://hf.co/hmbench/hmbench-ajmc-de-hmbyt5-bs8-wsFalse-e10-lr0.00016-poolingfirst-layers-1-crfFalse-2
[13]: https://hf.co/hmbench/hmbench-ajmc-de-hmbyt5-bs8-wsFalse-e10-lr0.00016-poolingfirst-layers-1-crfFalse-3
[14]: https://hf.co/hmbench/hmbench-ajmc-de-hmbyt5-bs8-wsFalse-e10-lr0.00016-poolingfirst-layers-1-crfFalse-4
[15]: https://hf.co/hmbench/hmbench-ajmc-de-hmbyt5-bs8-wsFalse-e10-lr0.00016-poolingfirst-layers-1-crfFalse-5
[16]: https://hf.co/hmbench/hmbench-ajmc-de-hmbyt5-bs8-wsFalse-e10-lr0.00015-poolingfirst-layers-1-crfFalse-1
[17]: https://hf.co/hmbench/hmbench-ajmc-de-hmbyt5-bs8-wsFalse-e10-lr0.00015-poolingfirst-layers-1-crfFalse-2
[18]: https://hf.co/hmbench/hmbench-ajmc-de-hmbyt5-bs8-wsFalse-e10-lr0.00015-poolingfirst-layers-1-crfFalse-3
[19]: https://hf.co/hmbench/hmbench-ajmc-de-hmbyt5-bs8-wsFalse-e10-lr0.00015-poolingfirst-layers-1-crfFalse-4
[20]: https://hf.co/hmbench/hmbench-ajmc-de-hmbyt5-bs8-wsFalse-e10-lr0.00015-poolingfirst-layers-1-crfFalse-5

The [training log](training.log) and TensorBoard logs are also uploaded to the model hub.

More information about fine-tuning can be found [here](https://github.com/stefan-it/hmBench).

# 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 ❤️