LLMs Project Collection
This repository collects short, focused demos of fine-tuning and using transformer models for common NLP tasks. Each folder contains a notebook, assets, and a small README with implementation notes.
Projects
- Sentiment (Tweets): 01_finetuned-BERT-sentiment-tweets : multiclass sentiment classification on Twitter (notebook + fine-tuned
bert-uncasedmodel). - Fake News (DistilBERT): 02_fake-news-detection-DistilBERT : fake-news detection demo using DistilBERT (notebook + model files).
- NER (Restaurant): 03_NER-recognition-DistilBERT : named entity recognition example for restaurant data (notebook).
- Resume Classification (BERT): 04_finetuned-BERT-resume-classification : classifies resumes into industries using a fine-tuned BERT model (notebook).
- Product Description (Phi-2): 05_finetuned-Phi2-generate-product-desc : generation demo fine-tuning Phi-2 for product descriptions (notebook).
Notes
- See each project's README for data sources, training settings, and reproduction notes.
- Notebooks were developed on Google Colab GPU; adjust device settings if running on CPU.
Questions or contributions: open an issue, submit a PR, or email afreenhossain0000@gmail.com.