The AI Almanac: tracking AI research through papers and code
Explore emerging AI research with The AI Almanac, powered by GitScholar: a dataset connecting papers, citations and GitHub engagement.
Keeping up with AI research takes more than scanning a list of new papers. The AI Almanac brings paper discovery and research trends together, combining academic information with activity around code on GitHub.
From papers to emerging ideas
Browse recent papers, explore research families and follow how ideas develop over time. The Almanac uses a graph neural network to estimate future citation attention from early signals, including citations, author history, paper content and repository activity. It also groups methods and ideas into a research taxonomy, with weekly data releases.
These scores help prioritize reading; they are not a verdict on scientific quality. Predictions for the newest papers carry the most uncertainty.
The research behind the Almanac
The GitScholar paper, released on arXiv on 22 September 2026, studies GitHub engagement as an early signal of research impact. Its dataset connects approximately 444,000 repositories with more than 558,000 AI papers, retaining timestamps so researchers can evaluate predictions against historical information.
Explore the public dataset on Hugging Face or visit the Almanac source repository.
