sentiment_fine_tune_bert

This model is a fine-tuned version of distilbert-base-uncased on a spam classification dataset. It achieves the following results on the evaluation set: {'eval_loss': 0.017569826330457415}

Intended uses & limitations

The model can be used for classifing whether the text is spam or not.

Training procedure

Trained using TFTrainer

Training hyperparameters

num_train_epochs = 2,
per_device_train_batch_size = 8,
per_device_eval_batch_size = 16,
eval_steps=100,
warmup_steps = 500,
weight_decay = 0.01,
logging_steps = 10,

Training results

Confusion matrix - [[955, 0], [ 0, 160]]

      precision    recall  f1-score   support

       0       1.00      1.00      1.00       955
       1       1.00      1.00      1.00       160

  accuracy                           1.00      1115
 macro avg       1.00      1.00      1.00      1115
weighted avg       1.00      1.00      1.00      1115

Framework versions

• Transformers 4.35.2
• TensorFlow 2.15.0
• Tokenizers 0.15.0