Edit model card

bert-uncased-keyword-discriminator

This model is a fine-tuned version of bert-base-uncased on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1296
  • Precision: 0.8439
  • Recall: 0.8722
  • Accuracy: 0.9727
  • F1: 0.8578
  • Ent/precision: 0.8723
  • Ent/accuracy: 0.9077
  • Ent/f1: 0.8896
  • Con/precision: 0.8010
  • Con/accuracy: 0.8196
  • Con/f1: 0.8102

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: 16
  • eval_batch_size: 16
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 8
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Precision Recall Accuracy F1 Ent/precision Ent/accuracy Ent/f1 Con/precision Con/accuracy Con/f1
0.1849 1.0 1875 0.1323 0.7039 0.7428 0.9488 0.7228 0.7522 0.8166 0.7831 0.6268 0.6332 0.6300
0.1357 2.0 3750 0.1132 0.7581 0.8024 0.9592 0.7796 0.7948 0.8785 0.8346 0.6971 0.6895 0.6933
0.0965 3.0 5625 0.1033 0.8086 0.7980 0.9646 0.8032 0.8410 0.8592 0.8500 0.7560 0.7071 0.7307
0.0713 4.0 7500 0.1040 0.8181 0.8435 0.9683 0.8306 0.8526 0.8906 0.8712 0.7652 0.7736 0.7694
0.0525 5.0 9375 0.1126 0.8150 0.8633 0.9689 0.8385 0.8495 0.9064 0.8770 0.7629 0.7993 0.7807
0.0386 6.0 11250 0.1183 0.8374 0.8678 0.9719 0.8523 0.8709 0.9020 0.8862 0.7877 0.8170 0.8021
0.03 7.0 13125 0.1237 0.8369 0.8707 0.9723 0.8535 0.8657 0.9079 0.8863 0.7934 0.8155 0.8043
0.0244 8.0 15000 0.1296 0.8439 0.8722 0.9727 0.8578 0.8723 0.9077 0.8896 0.8010 0.8196 0.8102

Framework versions

  • Transformers 4.19.2
  • Pytorch 1.11.0+cu113
  • Datasets 2.2.2
  • Tokenizers 0.12.1
Downloads last month
27
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.