output
This model is a fine-tuned version of PlanTL-GOB-ES/bsc-bio-ehr-es on the Rodrigo1771/distemist-ner dataset. It achieves the following results on the evaluation set:
- Loss: 0.1470
- Precision: 0.8067
- Recall: 0.8067
- F1: 0.8067
- Accuracy: 0.9762
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: 5e-05
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10.0
Training results
Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
---|---|---|---|---|---|---|---|
0.0833 | 0.9997 | 1701 | 0.0842 | 0.7302 | 0.7363 | 0.7332 | 0.9714 |
0.0549 | 2.0 | 3403 | 0.0859 | 0.7831 | 0.7948 | 0.7889 | 0.9748 |
0.0358 | 2.9997 | 5104 | 0.0885 | 0.7895 | 0.7995 | 0.7945 | 0.9759 |
0.0213 | 4.0 | 6806 | 0.1097 | 0.7689 | 0.8065 | 0.7873 | 0.9740 |
0.0128 | 4.9997 | 8507 | 0.1217 | 0.7921 | 0.7890 | 0.7905 | 0.9750 |
0.0101 | 6.0 | 10209 | 0.1480 | 0.7989 | 0.8140 | 0.8064 | 0.9760 |
0.0058 | 6.9997 | 11910 | 0.1470 | 0.8067 | 0.8067 | 0.8067 | 0.9762 |
0.0045 | 8.0 | 13612 | 0.1539 | 0.7978 | 0.8149 | 0.8063 | 0.9761 |
0.0024 | 8.9997 | 15313 | 0.1659 | 0.7939 | 0.8109 | 0.8023 | 0.9757 |
0.0012 | 9.9971 | 17010 | 0.1725 | 0.7958 | 0.8135 | 0.8046 | 0.9761 |
Framework versions
- Transformers 4.40.2
- Pytorch 2.2.1+cu121
- Datasets 2.19.1
- Tokenizers 0.19.1
- Downloads last month
- 11
Finetuned from
Dataset used to train Rodrigo1771/bsc-bio-ehr-es-distemist-ner
Evaluation results
- Precision on Rodrigo1771/distemist-nervalidation set self-reported0.807
- Recall on Rodrigo1771/distemist-nervalidation set self-reported0.807
- F1 on Rodrigo1771/distemist-nervalidation set self-reported0.807
- Accuracy on Rodrigo1771/distemist-nervalidation set self-reported0.976