isy503-a03

This model is a fine-tuned version of distilbert/distilbert-base-uncased on the IMDB Dataset of 50K Movie Reviews dataset. It achieves the following results on the evaluation set:

• Loss: 0.2328
• Accuracy: 0.9318

Model description

A sentiment analysis model used on a academic excercise to learn and practice Sentiment Analysis using DistilBERT.

Intended uses & limitations

It is only an academic excercise, which aims to be the foundation for other excercises such as improving the mdoel using multilanguage processing and multi-feature output (Likert Scale to improve output accuracy, rather than only POSITIVE and NEGATIVE)

Training and evaluation data

The training has been done using the following tutorial: Hugging Face: Text classification. And the evaluation has been done with a random sample of Movie and Amazon Product reviews.

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: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.2251	1.0	1563	0.2154	0.9189
0.1463	2.0	3126	0.2328	0.9318

Framework versions

• Transformers 4.40.1
• Pytorch 2.2.1+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1