Tülu 3
AIMade by AIWritten by AI (Claude) from the source below. Check anything important against the source.How it's made
Tülu 3 — the full recipe released, running from SFT to DPO to RLVR
Why it matters
The turning point from preference learning to verifiable rewards
Source — Primary sources first; secondary reports are labelled “Press”
Paperarxiv.orghttps://arxiv.org/abs/2411.15124Added in Week 35, 2026 · See that week
On the map — ⑫ Training Data line
- Previous stopCan AI scaling continue through 2030?
- This recordTülu 3
- Next stopNous Research releases DeMo paper and code; 15B model trained with DisTrO