<!-- 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. -->
bert-finetuned-ner-para
This model is a fine-tuned version of bert-base-cased on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.0818
- Precision: 0.6445
- Recall: 0.7434
- F1: 0.6904
- 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: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 4
Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
|---|---|---|---|---|---|---|---|
| No log | 1.0 | 80 | 0.2387 | 0.4214 | 0.4053 | 0.4132 | 0.9387 |
| No log | 2.0 | 160 | 0.1301 | 0.6105 | 0.6163 | 0.6134 | 0.9650 |
| No log | 3.0 | 240 | 0.0952 | 0.6515 | 0.7218 | 0.6849 | 0.9741 |
| No log | 4.0 | 320 | 0.0818 | 0.6445 | 0.7434 | 0.6904 | 0.9762 |
Framework versions
- Transformers 4.23.1
- Pytorch 1.12.1+cu113
- Datasets 2.6.1
- Tokenizers 0.13.1
