🚟 New blog post: On "infinite" learning-rate schedules and how to construct them from one checkpoint to the next fabian-sp.github.io/posts/2025/09/…

Long in the making, finally released: Apertus-8B and Apertus-70B, trained on 15T tokens of open data from over 1800 languages. Unique opportunity in academia to work on and train LLMs across the full-stack. We managed to pull off a pretraining run with some fun innovations, ...

Die Berliner Verkehrssenatorin Uta Bonde (CDU) jetzt im Gespräch mit dem Tagesspiegel zum Thema Schulwegsicherheit: „Wir können nicht nach Gutdünken Tempo 30 einführen“ Autofreie Schulstraßen wie in Paris sieht sie skeptisch. Dafür ihr Rat an alle Kinder und Jugendlichen: „Helm

what a beauty

If you’re scrambling a last-minute submission with an uncertain result, remember: putting it off is hard in the moment. It will sting for 10 minutes (because you care so deeply), but in 10 months you’ll be incredibly proud you made the scientifically rigorous call.

are models getting nervous when they are set from .train() to .eval()?

weight decay seems to be a hot topic of this year's ICLR submissions 👀

when the paper title is a question, you can usually guess the "answer"

most stylish theatre i've been to. don't miss the coffee bar in the break.

TIL: Even the Sophia authors couldn't reproduce the Sophia paper's results. source: arxiv.org/pdf/2509.02046

Good to see SOAP and Muon being quite performant in another setting — training of Diffusion Models. Similarly to our benchmark, the authors find Prodigy a decent “proxy-optimizer” for tuning hyperparams of Adam-like methods arxiv.org/pdf/2510.19376