AG News Embedding Logistic Regression Classifier
This model is a news topic classifier created for an Information Retrieval assignment.
Task
The task is to classify short news texts into one of four topic categories:
- World
- Sports
- Business
- Sci/Tech
Model Details
- Embedding model:
sentence-transformers/all-MiniLM-L6-v2
- Classifier: Logistic Regression
- Task: AG News topic classification
- Labels:
- 1 = World
- 2 = Sports
- 3 = Business
- 4 = Sci/Tech
- Training data: processed balanced AG News subset
Method
Each news text is first converted into a sentence embedding using sentence-transformers/all-MiniLM-L6-v2. A Logistic Regression classifier is then trained on these embeddings to predict the news topic.
Evaluation
The model was evaluated on a balanced test set of 1000 examples.
- Accuracy: 0.8830
- Macro F1: 0.8834
Files
classifier.joblib: trained Logistic Regression classifierlabel_map.json: mapping from numeric labels to readable class names
