A hands-on collection of Jupyter notebooks and Python examples that walk through the most popular NLP tasks—from classical pattern-matching with regular expressions to modern, parameter-efficient fine-tuning of transformer models using LoRA adapters. This repository might be helpful for those just getting started with NLP, offering useful tutorials and end-to-end pipelines. The content of the repository is summarized as follows:
- Word2Vec from Scratch: Implement the skip-gram model to train our own word vectors. Evaluate the model, and compare it with the existing LLMs.
- Stable Diffusion with LoRA: Fine-tune a Stable Diffusion model using low-rank adapters for efficient image generation. As a result, we can adapt the image generation model to generate images based on specific style or theme.
- Modern NLP Techniques: Hands-on examples of transformer-based tasks such as prompt engineering, architecture analysis and modification.
- Overview of other popular NLP tasks such as sentiment analysis, NER, assessing the quality of text embedding models, and more.