Shifting Focus from Optimizers to Weight Transformation
torchcompiled · x · 2026-07-15
The author reflects on previous attempts to modify things at the optimizer level, which often led to divergence, especially when trying to alter both per-dimension step sizes and preconditioning simultaneously.
Thus, the approach shifted to:
- Leaving the preconditioner untouched;
- Transforming the weights themselves instead;
- Introducing a scaling mechanism tied to parameter magnitude, aligning the "sense of distance" for parameter updates with actual optimization needs.
More from Research
- Diamandis: AI Cancer Navigator and $100 Genome Sequencing Close the Care Gap — PeterDiamandis · 2026-10-27
- EMNLP Oral: training-free Nexus Sampling cuts KV-cache memory 10x; OpenResearcher tops BrowseComp-Plus at 54.8% — yuz9yuz · 2026-10-27
- ICML 2026 position paper: LLM 'thinking traces' are a harmful anthropomorphism — rao2z · 2026-10-27
- Forking gap follows a power law in context frequency; masking rare n-grams shrinks it — ZimingLiu11 · 2026-10-27
- How forking happens: n-gram tables split logits into non-interacting subspaces — ZimingLiu11 · 2026-10-27
- The best LoRA explainer video yet: rank constraints meet intrinsic dimension — sophiamyang · 2026-10-27