Authors L\'eo Grinsztajn (Liam), Klemens Fl\"oge (Liam), Oscar Key (Liam), Felix Birkel (Liam), Philipp Jund (Liam), Brendan Roof (Liam), Mihir Manium (Liam), Shi Bin (Liam), Hoo, Magnus B\"uhler, Anurag Garg, Dominik Safaric, Jake Robertson, Benjamin J\"ager, Simone Alessi, Adrian Hayler, Vladyslav Moroshan, Lennart Purucker, Philipp Singer, Alan Arazi, Julien Siems, Jan Hendrik Metzen, Georg Grab, Nick Erickson, Siyuan Guo, Eliott Kalfon, Simon Bing, David Salinas, Clara Cornu, Lilly Charlotte Wehrhahn, Diana Kriuchkova, Kursat Kaya, Lydia Sidhoum, Marie Salmon, Jerry Chen, Madelon Hulsebos, Yann LeCun, Samuel M\"uller, Bernhard Sch\"olkopf, Sauraj Gambhir, Noah Hollmann, Frank Hutter
Released 2026-05-15 (2mo ago) Topic Relation to my work Read next because TabPFN-3: Technical Report overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, class, under, line, without, trained, test, model. Source: arxiv stat.ML (Machine Learning).
arXiv 2605.13986 Ask Claude about this paper Posts a comment with @claude so the runner spins up a Claude Code session scoped to TabPFN-3: Technical Report . Replies appear in the conversation below.
Summarize the contribution in 3 bullets. How does this relate to my current research? What experiments would test the central claim? What are the weakest assumptions in the paper?