Downloads · 30 days
5
7% of all-time downloads
cross-lingual-transfer-demo-ner-2
zzzotop/cross-lingual-transfer-demo-ner-2
cross-lingual-transfer-demo-ner-2 is a token classification model from zzzotop. Use it when you need labels on individual words, such as names. It is set up for transformers.
Cross-lingual transfer demo with Faroese named entity recognition. Model is trained on 104 languages, including Norwegian, Danish, Swedish, and Icelandic (but not Faroese). Based on 'Transfer to a Low-Resource Languag…
All-time downloads
71
Public
Repo size
1.4 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.bin709 MB · 99%
At a glance
- Task
- Token Classification
- Library
- transformers
- Model type
- bert
- Access
- Public
- Created
- Aug 27, 2023
- Updated
- Sep 4, 2023
- SHA
- b40e3b33
Try a prompt
Model card
From the Hugging Face model README
Cross-lingual transfer demo with Faroese named entity recognition. Model is trained on 104 languages, including Norwegian, Danish, Swedish, and Icelandic (but not Faroese). Based on 'Transfer to a Low-Resource Language via Close Relatives: The Case Study on Faroese', Snæbjarnarson et al. (2023).
bert-base-multilingual-cased finetuned on vesteinn/sosialurin-faroese-ner. Trained for 5 epochs, 22385 steps, lr 2e-5.<b> Training loop written in native pytorch.</b>
Evaluation over all batches, methodology inspired by seqeval: <br> Accuracy: 0.48326 <br> Precision: 0.48326 (micro) 0.53593 (macro) 0.24632 (weighted) <br> Recall: 0.48326 (micro) 0.78419 (macro) 0.48326 (weighted) <br> F1: 0.48326 (micro) 0.60675 (macro) 0.32438 (weighted)
- Task
- Token Classification
- Library
- transformers
- Type
- bert
- Created
- Aug 27, 2023
- Updated
- Sep 4, 2023