TTME beats test-time training on 5 math tasks while cutting FLOPs by 82.8-84.6%
UCSantaBarbara · hf · 2026-10-28
Researchers from UC Santa Barbara and UC Berkeley propose Test-Time Memory Evolution (TTME), a plug-and-play alternative to test-time training (TTT) for scientific discovery. Instead of costly online weight updates, TTME continually distills experience into external memory to guide subsequent discovery. Under an identical discovery framework with only the adaptation method swapped, TTME outperforms TTT on all five mathematical tasks while reducing computational cost by 82.8-84.6% in FLOPs.
More from Research
- Ben Lorica: coding agents that build the robot instead of controlling it — bigdata · 2026-10-28
- RAW-RAG Open-Sourced: 256-Token Windows + Rerank Baseline Challenges Complex Agentic Memory — JeremyCMorgan · 2026-10-28
- Open-source repo shares 20 exploratory NCM and GPT-2 interpretability studies — BedFalse6021 · 2026-10-28
- Can you overthink thinking itself? Russell & Wefald's 90s metareasoning work says the recursion never bottoms out — BraydonDymm · 2026-10-28
- Generate Biomed's AI-designed CAR-T scaffold is 70% smaller with 3x potency — kenbwork · 2026-10-28
- Linear probe predicts VGGT pose failures at 93.5% AUC, paper finds — kwangmoo_yi · 2026-10-28