1. Say what the post is about
State the main point early. Tell the reader what you found, built, tested, or argued, and why it matters for Apertus.
We welcome submissions from students, independent researchers and experts across disciplines. You do not need to submit a large or fully polished result — what matters is a clear contribution that helps others understand Apertus better. That can be a technical analysis, a useful tool, a conceptual idea, or a well-documented codebase. Good submissions usually include something concrete the reader can inspect, such as figures, diagrams, examples, experiments, interactive artifacts, or code. The strongest posts make their point early, explain why the work matters, and give enough detail for others to follow, reuse, or build on it. Every submission is reviewed by the Claritas review team, but everyone is welcome to contribute. We want the site to reflect an active interpretability community around Apertus, where many people can add something useful, even if the work is exploratory or only partially developed.
State the main point early. Tell the reader what you found, built, tested, or argued, and why it matters for Apertus.
Choose the artifact type that best fits the submission. If you are not sure, pick the closest match and explain it in the TLDR or opening paragraph.
Use a clear structure, short sections, and direct language. If the post depends on results, include the setup, the model or checkpoint used, and the key details someone would need to follow the work.
Do not make the claim sound bigger than the support behind it. If the result is early, partial, or exploratory, say that plainly. If it is stronger or more complete, show the evidence cleanly.
Add links to code, datasets, demos, or references when they help the reader inspect or reuse the work. Add an image, figure, or diagram if it makes the submission easier to understand.