Downloads · 30 days
13
3% of all-time downloads
hmbert-tiny/flair-icdar-fr
flair-icdar-fr is a token classification model from hmbert-tiny. Use it when you need labels on individual words, such as names. It is set up for flair. The card lists the license as mit.
This Flair model was fine-tuned on the French ICDAR-Europeana NER Dataset using hmBERT Tiny as backbone LM.
Downloads · 30 days
13
3% of all-time downloads
All-time downloads
501
Public
Repo size
20.2 MB
Likes
0
Public
Click a slice to open those files.
.bin19 MB · 92%
From the Hugging Face model README
This Flair model was fine-tuned on the French ICDAR-Europeana NER Dataset using hmBERT Tiny as backbone LM.
The ICDAR-Europeana NER Dataset is a preprocessed variant of the Europeana NER Corpora for Dutch and French.
The following NEs were annotated: PER, LOC and ORG.
We performed a hyper-parameter search over the following parameters with 5 different seeds per configuration:
[4, 8][5e-05, 3e-05]And report micro F1-score on development set:
| Configuration | Seed 1 | Seed 2 | Seed 3 | Seed 4 | Seed 5 | Average |
|---|---|---|---|---|---|---|
bs4-e10-lr5e-05 | 0.6013 | 0.5273 | 0.6086 | 0.6208 | 0.5731 | 0.5862 ± 0.0373 |
bs8-e10-lr5e-05 | 0.6186 | 0.4917 | 0.6056 | 0.5972 | 0.4881 | 0.5602 ± 0.0647 |
bs4-e10-lr3e-05 | 0.6034 | 0.4735 | 0.5837 | 0.578 | 0.4716 | 0.542 ± 0.0641 |
bs8-e10-lr3e-05 | 0.5743 | 0.4119 | 0.551 | 0.5261 | 0.4408 | 0.5008 ± 0.0708 |
The 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.
We thank Luisa März, Katharina Schmid and Erion Çano for their fruitful discussions about Historic Language Models.
Research supported with Cloud TPUs from Google's TPU Research Cloud (TRC). Many Thanks for providing access to the TPUs ❤️