my-sentiment-model

A binary sentiment classification model that labels English text as POSITIVE or NEGATIVE. It is a fine-tuned version of distilbert-base-uncased trained on the IMDB movie reviews dataset.

Model description

The model takes a piece of English text and predicts whether the overall sentiment is positive or negative. It was built by fine-tuning DistilBERT — a smaller, faster distilled version of BERT — on labelled movie reviews, making it lightweight enough to run on CPU while retaining strong accuracy.

• Base model: distilbert-base-uncased
• Task: Binary sentiment classification (text-classification)
• Language: English
• Labels: NEGATIVE (0), POSITIVE (1)

Intended uses & limitations

Intended uses: Quick sentiment scoring of English text such as reviews, comments, and short feedback.

Limitations:

• English only — performance on other languages is not reliable.
• Trained on movie reviews, so accuracy is highest on review-style text and may drop on very different domains (e.g. legal, medical, or social media slang).
• Binary output only; it does not detect neutral sentiment, sarcasm, or mixed opinions well.

How to use

from transformers import pipeline

clf = pipeline("text-classification", model="rajaadil/my-sentiment-model")
print(clf("This movie was absolutely fantastic!"))
# [{'label': 'POSITIVE', 'score': 0.98}]

Or load the model and tokenizer directly:

from transformers import AutoTokenizer, AutoModelForSequenceClassification

tokenizer = AutoTokenizer.from_pretrained("rajaadil/my-sentiment-model")
model = AutoModelForSequenceClassification.from_pretrained("rajaadil/my-sentiment-model")

Training data

The model was trained on a subset of the IMDB dataset:

• Training set: 2,000 reviews
• Evaluation set: 1,000 reviews

Training procedure

Evaluation results

Framework versions

• Transformers
• PyTorch
• Datasets

Author

Built by rajaadil.