changwookjun/nlp-paper

What it solves

It addresses the challenge of keeping up with the rapidly evolving NLP research landscape by gathering influential papers into one organized reference. Instead of searching across multiple sources, researchers can quickly find key works on specific subtopics.

How it works

The repository is a simple curated list. Papers are grouped under thematic headings like "Bert Series", "Transformer Series", "Transfer Learning", "Question Answering", and more recent categories like "LLM", "Retrieval Augmented Generation", and "Safety Hallucination and Interpretability". Each entry links to the paper's arXiv or other source, and some include a link to a GitHub implementation. The list is maintained as a markdown file with a table of contents for navigation.

Who it's for

This is for NLP researchers, students, and practitioners who want a quick overview of important papers in a given area, or who need a reference for building their own reading lists. It's also useful for those tracking the evolution of methods from BERT-era models to modern large language models.

Highlights

• Covers classic models like BERT and Transformer variants as well as recent LLM topics such as alignment, RAG, and agents.
• Papers are annotated with venue and year where available, helping readers gauge the significance and timeline.
• Includes a mix of foundational theory, analysis, and practical fine-tuning methods (e.g., LoRA, QLoRA).
• Many entries link to open-source implementations, enabling hands-on exploration.