A better metaphor for LLMs: bags of contextually activated circuits and heuristics
xuanalogue · x · 2026-09-27
xuanalogue wishes for a better metaphor for how LLMs work at the mechanistic level, describing them as "bags of contextually activated circuits, heuristics, and algorithms" — echoing interpretability research that views models as many dynamically activated small circuits rather than one monolithic algorithm.
Related event: LLMs as Bags of Contextually Activated Heuristics(3 posts)→
More from Research
- kalomaze proposes testing which nanogpt tricks survive causal NTP over DCT coefficients — kalomaze · 2026-09-27
- AI-designed drug turns back 6 aging clocks; semaglutide extends mouse lifespan 12% — rand_longevity · 2026-09-27
- Could perfect monosemanticity enable training coding agents without verifiers or RL? — menhguin · 2026-09-27
- Local agent eval: challenger model wrote tool calls as prose 5/80 times — a silent failure class — Grimmoner · 2026-09-27
- TalkPlayData-backed conversational music recsys challenge at RecSys 2026 draws 41 teams — keunwoochoi · 2026-09-27
- AgentSeism: open-source statistical CI for deciding when an agent truly regressed — puppy_lover_2021 · 2026-09-27