paper_notes
Daily notes on AI research, especially those on LLMs, VLMs, AI agents and related techniques such as pre-, mid-, post-training, RL, continual learning, test-tim
What is paper_notes?
Daily notes on AI research, especially those on LLMs, VLMs, AI agents and related techniques such as pre-, mid-, post-training, RL, continual learning, test-time scaling, RSI, and more.
Overview of paper_notes
What it is
paper_notes is a human-curated collection of AI research papers and related resources, including blog posts. The maintainer follows discussions on X daily to keep up with developments in AI. Each item is managed as a GitHub Issue. All content stored in GitHub Issues is exported daily as XML files under agent_docs/. The primary purpose of this repository is to serve as a personal study log and memory aid for the maintainer.
Key features
- Curated from what is being discussed on X, focusing on papers and resources that attract broad interest.
- Tagged for discovery: manually assigned tags reflect the maintainer's interpretation of each item.
- Notes based on the maintainer's understanding, with some items including notes in Japanese.
- Paper metadata and abstracts are retrieved via the arXiv API.
- Japanese translations and summaries of abstracts are automatically generated using GPT.
- Over 600 different labels exist for filtering issues.
- A Wiki is maintained to record personal impressions of trends observed while browsing papers and discussions on X.
Use cases
- Keeping a personal study log and memory aid for AI research.
- Exploring AI research papers and resources curated from discussions on X.
- Using an AI agent to answer questions across the collection via the exported XML files.
- Interactively extracting information from issues and notes via DeepWiki/CodeWiki chat-based interfaces.
Getting started
Clone the repository locally and use an agent to chat with the repository based on the contents of its issues.bashgit clone https://github.com/AkihikoWatanabe/paper_notes.gitcd paper_notes
License and pricing
License: MIT. Pricing type: open source.
Based on the project's public README and repository metadata.