ADANA outscales AdamW across overtraining in 51M–253M Transformers
_katieeverett · x · 2026-09-10
Core result of the thread: across 51M–253M parameter Transformers, ADANA outscales AdamW as the overtraining factor grows. Matrix-preconditioned optimizers lead at small OT; Muon's advantage is roughly constant across OT, while SOAP may gain further at the highest OT factors tested.
More from Research
- Researchers identify "Forking": train loss falls while validation loss turns up — ZimingLiu11 · 2026-10-26
- Google DeepMind pilots world's first double-blind evaluations for frontier models — scychan_brains · 2026-10-26
- Rewriting scaling law study: optimal synthetic data share rises from 10% to 32% with compute — heghbalz · 2026-10-26
- FLASH: GPU-native deformable sim learns cloth folding in minutes on one RTX 5090 — ChongZzZhang · 2026-10-26
- LEMON: zero-shot LLM misbehavior monitor flags hallucinations from attention maps — yoavartzi · 2026-10-26
- Behind the EMNLP keynote: LMs as simulated learners to generate child language acquisition hypotheses — najoungkim · 2026-10-26