File size: 5,003 Bytes
9a6ffa2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
language: fr
license: mit
tags:
- flair
- token-classification
- sequence-tagger-model
base_model: dbmdz/bert-base-historic-multilingual-64k-td-cased
widget:
- text: Nous recevons le premier numéro d ' un nouveau journal , le Radical - Libéral
    , qui paraîtra à Genève deux fois la semaine . Son but est de représenter l '
    élément national du radicalisme genevois , en d ' autres termes , de défendre
    la politique intransigeante do M . Carteret , en opposition aux tendances du groupe
    _ > dont le Genevois est l ' organe . Bétail .
---

# Fine-tuned Flair Model on French HIPE-2020 Dataset (HIPE-2022)

This Flair model was fine-tuned on the
[French HIPE-2020](https://github.com/hipe-eval/HIPE-2022-data/blob/main/documentation/README-hipe2020.md)
NER Dataset using hmBERT 64k as backbone LM.

The HIPE-2020 dataset is comprised of newspapers from mid 19C to mid 20C. For information can be found
[here](https://dl.acm.org/doi/abs/10.1007/978-3-030-58219-7_21).

The following NEs were annotated: `loc`, `org`, `pers`, `prod`, `time` and `comp`.

# Results

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

* Batch Sizes: `[4, 8]`
* Learning Rates: `[3e-05, 5e-05]`

And report micro F1-score on development set:

| Configuration     | Seed 1       | Seed 2       | Seed 3           | Seed 4       | Seed 5       | Average         |
|-------------------|--------------|--------------|------------------|--------------|--------------|-----------------|
| `bs8-e10-lr3e-05` | [0.8389][1]  | [0.8466][2]  | [0.8299][3]      | [0.8391][4]  | [0.8427][5]  | 0.8394 ± 0.0062 |
| `bs4-e10-lr3e-05` | [0.8279][6]  | [0.8364][7]  | [0.8404][8]      | [0.8382][9]  | [0.8371][10] | 0.836 ± 0.0048  |
| `bs8-e10-lr5e-05` | [0.8418][11] | [0.8337][12] | [0.831][13]      | [0.8346][14] | [0.8352][15] | 0.8353 ± 0.004  |
| `bs4-e10-lr5e-05` | [0.831][16]  | [0.8239][17] | [**0.7784**][18] | [0.8313][19] | [0.8191][20] | 0.8167 ± 0.022  |

[1]: https://hf.co/stefan-it/hmbench-hipe2020-fr-hmbert_64k-bs8-wsFalse-e10-lr3e-05-poolingfirst-layers-1-crfFalse-1
[2]: https://hf.co/stefan-it/hmbench-hipe2020-fr-hmbert_64k-bs8-wsFalse-e10-lr3e-05-poolingfirst-layers-1-crfFalse-2
[3]: https://hf.co/stefan-it/hmbench-hipe2020-fr-hmbert_64k-bs8-wsFalse-e10-lr3e-05-poolingfirst-layers-1-crfFalse-3
[4]: https://hf.co/stefan-it/hmbench-hipe2020-fr-hmbert_64k-bs8-wsFalse-e10-lr3e-05-poolingfirst-layers-1-crfFalse-4
[5]: https://hf.co/stefan-it/hmbench-hipe2020-fr-hmbert_64k-bs8-wsFalse-e10-lr3e-05-poolingfirst-layers-1-crfFalse-5
[6]: https://hf.co/stefan-it/hmbench-hipe2020-fr-hmbert_64k-bs4-wsFalse-e10-lr3e-05-poolingfirst-layers-1-crfFalse-1
[7]: https://hf.co/stefan-it/hmbench-hipe2020-fr-hmbert_64k-bs4-wsFalse-e10-lr3e-05-poolingfirst-layers-1-crfFalse-2
[8]: https://hf.co/stefan-it/hmbench-hipe2020-fr-hmbert_64k-bs4-wsFalse-e10-lr3e-05-poolingfirst-layers-1-crfFalse-3
[9]: https://hf.co/stefan-it/hmbench-hipe2020-fr-hmbert_64k-bs4-wsFalse-e10-lr3e-05-poolingfirst-layers-1-crfFalse-4
[10]: https://hf.co/stefan-it/hmbench-hipe2020-fr-hmbert_64k-bs4-wsFalse-e10-lr3e-05-poolingfirst-layers-1-crfFalse-5
[11]: https://hf.co/stefan-it/hmbench-hipe2020-fr-hmbert_64k-bs8-wsFalse-e10-lr5e-05-poolingfirst-layers-1-crfFalse-1
[12]: https://hf.co/stefan-it/hmbench-hipe2020-fr-hmbert_64k-bs8-wsFalse-e10-lr5e-05-poolingfirst-layers-1-crfFalse-2
[13]: https://hf.co/stefan-it/hmbench-hipe2020-fr-hmbert_64k-bs8-wsFalse-e10-lr5e-05-poolingfirst-layers-1-crfFalse-3
[14]: https://hf.co/stefan-it/hmbench-hipe2020-fr-hmbert_64k-bs8-wsFalse-e10-lr5e-05-poolingfirst-layers-1-crfFalse-4
[15]: https://hf.co/stefan-it/hmbench-hipe2020-fr-hmbert_64k-bs8-wsFalse-e10-lr5e-05-poolingfirst-layers-1-crfFalse-5
[16]: https://hf.co/stefan-it/hmbench-hipe2020-fr-hmbert_64k-bs4-wsFalse-e10-lr5e-05-poolingfirst-layers-1-crfFalse-1
[17]: https://hf.co/stefan-it/hmbench-hipe2020-fr-hmbert_64k-bs4-wsFalse-e10-lr5e-05-poolingfirst-layers-1-crfFalse-2
[18]: https://hf.co/stefan-it/hmbench-hipe2020-fr-hmbert_64k-bs4-wsFalse-e10-lr5e-05-poolingfirst-layers-1-crfFalse-3
[19]: https://hf.co/stefan-it/hmbench-hipe2020-fr-hmbert_64k-bs4-wsFalse-e10-lr5e-05-poolingfirst-layers-1-crfFalse-4
[20]: https://hf.co/stefan-it/hmbench-hipe2020-fr-hmbert_64k-bs4-wsFalse-e10-lr5e-05-poolingfirst-layers-1-crfFalse-5

The [training log](training.log) and TensorBoard logs (not available for hmBERT Base model) 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 ❤️