Downloads · 30 days
8
17% of all-time downloads
COMPner2-bert-base-spanish-wwm-cased
simonestradasch/COMPner2-bert-base-spanish-wwm-cased
COMPner2-bert-base-spanish-wwm-cased is a token classification model from simonestradasch. Use it when you need labels on individual words, such as names. It is set up for transformers.
should probably proofread and complete it, then remove this comment. --
All-time downloads
48
Public
Parameters
109M
437 MB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors437 MB · 100%
Parameter types
How the weights are stored.
F32109M · 100%
Try a prompt
Datasets
Model card
From the Hugging Face model README
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
COMPner2-bert-base-spanish-wwm-cased
This model is a fine-tuned version of dccuchile/bert-base-spanish-wwm-cased on the simonestradasch/NERcomp2 dataset. It achieves the following results on the evaluation set:
- Loss: 0.2843
- Body Part Precision: 0.6644
- Body Part Recall: 0.7143
- Body Part F1: 0.6884
- Body Part Number: 413
- Disease Precision: 0.7251
- Disease Recall: 0.7303
- Disease F1: 0.7276
- Disease Number: 975
- Family Member Precision: 0.8065
- Family Member Recall: 0.8333
- Family Member F1: 0.8197
- Family Member Number: 30
- Medication Precision: 0.7778
- Medication Recall: 0.6774
- Medication F1: 0.7241
- Medication Number: 93
- Procedure Precision: 0.5763
- Procedure Recall: 0.5949
- Procedure F1: 0.5854
- Procedure Number: 311
- Overall Precision: 0.6885
- Overall Recall: 0.7025
- Overall F1: 0.6955
- Overall Accuracy: 0.9146
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 13
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 2
Training results
| Training Loss | Epoch | Step | Validation Loss | Body Part Precision | Body Part Recall | Body Part F1 | Body Part Number | Disease Precision | Disease Recall | Disease F1 | Disease Number | Family Member Precision | Family Member Recall | Family Member F1 | Family Member Number | Medication Precision | Medication Recall | Medication F1 | Medication Number | Procedure Precision | Procedure Recall | Procedure F1 | Procedure Number | Overall Precision | Overall Recall | Overall F1 | Overall Accuracy |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.4243 | 1.0 | 1004 | 0.2935 | 0.5910 | 0.6998 | 0.6408 | 413 | 0.6784 | 0.6944 | 0.6863 | 975 | 0.8 | 0.8 | 0.8000 | 30 | 0.6882 | 0.6882 | 0.6882 | 93 | 0.6050 | 0.5466 | 0.5743 | 311 | 0.6473 | 0.6718 | 0.6593 | 0.9052 |
| 0.2348 | 2.0 | 2008 | 0.2843 | 0.6644 | 0.7143 | 0.6884 | 413 | 0.7251 | 0.7303 | 0.7276 | 975 | 0.8065 | 0.8333 | 0.8197 | 30 | 0.7778 | 0.6774 | 0.7241 | 93 | 0.5763 | 0.5949 | 0.5854 | 311 | 0.6885 | 0.7025 | 0.6955 | 0.9146 |
Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu118
- Datasets 2.13.1
- Tokenizers 0.13.3
- Task
- Token Classification
- Library
- transformers
- Type
- bert
- Languages
- es
- Created
- Jul 10, 2023
- Updated
- Jul 10, 2023