24679-HW2-text-distilbert-predictor

24679-HW2-text-distilbert-predictor

This model is a fine-tuned version of distilbert-base-uncased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0008
• Accuracy: 1.0
• F1: 1.0
• Precision: 1.0
• Recall: 1.0

Model description

This model uses the dataset jennifee/HW1-aug-text-dataset which contains a text Review of a book, and classifies it as Fiction (1) or Nonfiction(0). The model seeks to fine-tune DistilBERT to accomplish this classification task. It computes a confusion matrix and a brief error analysis of several misclassifications.

Intended uses & limitations

This is intended to perform and demonstrate finetuning a text model. It is trained on a relatively small dataset and so may not be applicable for large scale applications.

Training and evaluation data

This dataset contains a text 'Review' of a book, and classifies it as Fiction (1) or Nonfiction(0) It contains two splits of original (100 rows) and augmented (1.6k rows) The augmented dataset is split into train/validation/test sets using an 80/20 split

Training procedure

The model is tokenized for binary classifiation such that train: 1024 | val: 256 | test: 320 | ext_valid: 100

The hypermaraters for training are listed below

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 5

Training results

Framework versions

• Transformers 4.56.1
• Pytorch 2.8.0+cu126
• Datasets 4.0.0
• Tokenizers 0.22.0