Downloads · 30 days
13
1% of all-time downloads
Nonzerophilip/bert-finetuned-ner_swedish_small_set_health_and_standart
bert-finetuned-ner_swedish_small_set_health_and_standart is a token classification model from Nonzerophilip. Use it when you need labels on individual words, such as names. It is set up for transformers.
This model is a fine-tuned version of KBLab/bert-base-swedish-cased-nerfor only Swedish. It has been fine-tuned on the concatenation of a smaller version of SUC 3.0 and some medical text from the Swedish website 1177.
Downloads · 30 days
13
1% of all-time downloads
All-time downloads
2K
Public
Repo size
2 GB
Likes
0
Public
Click a slice to open those files.
.bin497 MB · 100%
From the Hugging Face model README
This model is a fine-tuned version of KBLab/bert-base-swedish-cased-nerfor only Swedish. It has been fine-tuned on the concatenation of a smaller version of SUC 3.0 and some medical text from the Swedish website 1177.
The model will predict the following entities:
| Tag | Name | Exampel |
|---|---|---|
| PER | Person | (e.g., Johan and Sofia) |
| LOC | Location | (e.g., Göteborg and Spanien) |
| ORG | Organisation | (e.g., Volvo and Skatteverket) \ |
| PHARMA_DRUGS | Medication | (e.g., Paracetamol and Omeprazol) |
| HEALTH | Illness/Diseases | (e.g., Cancer, sjuk and diabetes) |
| Relation | Family members | (e.g., Mamma and Farmor) |
It achieves the following results on the evaluation set:
More information needed
More information needed
More information needed
The following hyperparameters were used during training:
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
|---|---|---|---|---|---|---|---|
| No log | 1.0 | 219 | 0.1123 | 0.7674 | 0.6567 | 0.7078 | 0.9681 |
| No log | 2.0 | 438 | 0.0934 | 0.7643 | 0.7662 | 0.7652 | 0.9738 |
| 0.1382 | 3.0 | 657 | 0.0963 | 0.7548 | 0.7811 | 0.7677 | 0.9756 |