What it does

This is a curated, categorized reading list of NLP papers from the transformer age onward. The maintainer, Changwook Jun, organizes links by topic—BERT variants, efficient transformers, transfer learning, summarization, sentiment analysis, and roughly two dozen other buckets including LLMs and model compression.

The interesting bit

The list is actively maintained with a Korean NLP practitioner’s eye: it tracks not just the famous papers but follow-up analyses (“What does BERT learn about the structure of language?”), probing tools, and implementation repos. The downstream task section is unusually granular, splitting into 13 subcategories from slot filling to commonsense reasoning.

Key highlights

• Covers 2017–2022+ era, from “Attention Is All You Need” through LoRA and Performers
• Each entry links directly to arXiv or PDF; many include GitHub repo links
• Sections on non-English models and domain-specific fine-tuning, often neglected in English-centric lists
• Includes survey papers and quality evaluators alongside task-specific work
• 595 stars suggests it’s found a steady audience, though the repo itself is just the README

Caveats

• No search, no tagging, no abstracts—just headings and links
• “Surver paper” typo and inconsistent formatting suggest one-person maintenance without automation
• Truncated README in the source view; full list may be longer than visible

Verdict

Useful if you want a human-filtered starting point for a new NLP subfield and don’t mind scrolling. Skip it if you need searchable metadata, BibTeX, or systematic coverage—Papers With Code or Semantic Scholar will serve you better for deep research.