What we publish
The site has two deliberately separate layers. Field guides are original, first-party explanations of durable concepts such as retrieval, agent loops, multimodal alignment, and efficient inference. Source notes are an automated index built from arXiv metadata. Source notes preserve paper titles, author names, abstracts, and primary-source links for discovery; they are labeled, excluded from search indexing, and are not represented as tutorials.
A paper appearing in the research index is not an endorsement. It means the paper matched a broad topic filter, not that its methods or claims were independently reproduced.
Why the distinction matters
Research abstracts are written by a paper’s authors to summarize their own work. They are useful starting points, but they do not supply independent judgment. A useful educational guide should add a mental model, identify assumptions, explain relevant baselines, and tell the reader what evidence to inspect. Paper2Tutorial now makes that threshold explicit.
Who this is for
The guides are written for software engineers, product builders, students, and curious practitioners. They assume basic familiarity with machine learning vocabulary but avoid requiring a graduate-level mathematical background. They are educational material, not professional, safety, medical, legal, or investment advice.
Corrections and accountability
AI research changes quickly and mistakes are possible. Corrections, attribution concerns, and source suggestions are welcome through the public route on the contact page. The standards for sourcing, automation, updates, and conflicts are documented in the editorial policy.