machineuser
Automatic correction of README.md metadata. Contact [email protected] for any question
a068117
metadata
tags:
  - generated_from_trainer
metrics:
  - precision
  - recall
  - f1
  - accuracy
model_index:
  - name: bert-srb-ner-setimes
    results:
      - task:
          name: Token Classification
          type: token-classification
        metric:
          name: Accuracy
          type: accuracy
          value: 0.9645112274185379

bert-srb-ner-setimes

This model was trained from scratch on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1955
  • Precision: 0.8229
  • Recall: 0.8465
  • F1: 0.8345
  • Accuracy: 0.9645

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: 32
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 20

Training results

Training Loss Epoch Step Validation Loss Precision Recall F1 Accuracy
No log 1.0 104 0.2281 0.6589 0.7001 0.6789 0.9350
No log 2.0 208 0.1833 0.7105 0.7694 0.7388 0.9470
No log 3.0 312 0.1573 0.7461 0.7778 0.7616 0.9525
No log 4.0 416 0.1489 0.7665 0.8091 0.7872 0.9557
0.1898 5.0 520 0.1445 0.7881 0.8327 0.8098 0.9587
0.1898 6.0 624 0.1473 0.7913 0.8316 0.8109 0.9601
0.1898 7.0 728 0.1558 0.8101 0.8347 0.8222 0.9620
0.1898 8.0 832 0.1616 0.8026 0.8302 0.8162 0.9612
0.1898 9.0 936 0.1716 0.8127 0.8409 0.8266 0.9631
0.0393 10.0 1040 0.1751 0.8140 0.8369 0.8253 0.9628
0.0393 11.0 1144 0.1775 0.8096 0.8420 0.8255 0.9626
0.0393 12.0 1248 0.1763 0.8161 0.8386 0.8272 0.9636
0.0393 13.0 1352 0.1949 0.8259 0.8400 0.8329 0.9634
0.0393 14.0 1456 0.1842 0.8205 0.8420 0.8311 0.9642
0.0111 15.0 1560 0.1862 0.8160 0.8493 0.8323 0.9646
0.0111 16.0 1664 0.1989 0.8176 0.8367 0.8270 0.9627
0.0111 17.0 1768 0.1945 0.8246 0.8409 0.8327 0.9638
0.0111 18.0 1872 0.1997 0.8270 0.8426 0.8347 0.9634
0.0111 19.0 1976 0.1917 0.8258 0.8491 0.8373 0.9651
0.0051 20.0 2080 0.1955 0.8229 0.8465 0.8345 0.9645

Framework versions

  • Transformers 4.9.2
  • Pytorch 1.9.0
  • Datasets 1.11.0
  • Tokenizers 0.10.1