Sakana AI Introduces Fast-Weight Product Key Memory Method
SakanaAILabs · x · 2026-07-04
Sakana AI's paper "Fast-weight Product Key Memory" (FwPKM) is slated for an oral presentation at the ICML 2026 "Towards Sustainable AI: Continuous Adaptation" workshop. The method combines Test-Time Training (TTT) with Product Key Memory (PKM) to create sparse fast weights, enabling high-capacity memory construction during inference without updating all weights. It offers large storage capacity, sparse access, and supports test-time writing. Under a 128K context length, it achieves about 75% accuracy on the 5-needle NIAH benchmark, and rapidly adapts to new domains after reading 4 million tokens from a target domain, with ΔNLL dropping by up to 0.4.
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