François Fleuret: why isn't EMA normalization more popular than batch norm?
francoisfleuret · x · 2026-11-13
Researcher François Fleuret asks why EMA normalization isn't more popular, arguing it makes far more sense than any batch-based normalization — "don't tell me it doesn't work."
The post is an open provocation about normalization design in training, contrasting batch-statistics approaches like BatchNorm with running-average alternatives.
More from Research
- OpenKB: open-source LLM knowledge base hits 4.9k stars with vectorless reasoning-based retrieval — tom_doerr · 2026-11-13
- PhysCaP: NTU and NVIDIA's agents measure mass and stiffness with no extra sensors — shaohua0116 · 2026-11-13
- METR time-horizon fit called too pessimistic, redone — with a superexponential meme chart — lxrjl · 2026-11-13
- Meta's Papailiopoulos: Training's Future Is Agents That Compile Intent Into Fully Bespoke FLOPs — DimitrisPapail · 2026-11-13
- Further fine-tuning CLIP with contrastive losses often fails on out-of-distribution data — wightmanr · 2026-11-13
- Interaction-as-a-Judge beats code rewards: +6.7pp win rate, 35.7% less code bloat — arena · 2026-11-13