Downloads · 30 days
3.6K
3% of all-time downloads
bert-base-swedish-lowermix-reallysimple-ner
KBLab/bert-base-swedish-lowermix-reallysimple-ner
bert-base-swedish-lowermix-reallysimple-ner is a token classification model from KBLab. Use it when you need labels on individual words, such as names. It is set up for transformers. The card lists the license as mit.
This model is based on KB-BERT and was fine-tuned on the SUCX 3.0 - NER corpus, using the simple tags and partially lowercased data. For this model we used a variation of the data that did not use BIO-encoding to diff…
All-time downloads
135K
Public
Parameters
124M
993 MB on disk
Likes
5
Public
Repo makeup
Click a slice to open those files.
.bin497 MB · 50%
Parameter types
How the weights are stored.
F32124M · 100%
Try a prompt
Datasets
Model card
From the Hugging Face model README
KB-BERT for NER
Mixed cased and uncased data
This model is based on KB-BERT and was fine-tuned on the SUCX 3.0 - NER corpus, using the simple tags and partially lowercased data. For this model we used a variation of the data that did not use BIO-encoding to differentiate between the beginnings (B), and insides (I) of named entity tags.
The model was trained on the training data only, with the best model chosen by its performance on the validation data. You find more information about the model and the performance on our blog: https://kb-labb.github.io/posts/2022-02-07-sucx3_ner
Dataset source
We have used a processed and remixed version of SUC 3.0 to train this model. See [SUCX 3.0][https://huggingface.co/datasets/KBLab/sucx3_ner].
Original dataset citation: Språkbanken (2024). SUC 3.0 (updated: 2024-06-03). [Data set]. Enriched and distributed by Språkbanken. https://doi.org/10.23695/wy84-ar30
Model license
MIT License
- Task
- Token Classification
- Library
- transformers
- Type
- bert
- License
- mit
- Languages
- sv
- Created
- Mar 2, 2022
- Updated
- Oct 15, 2026