attention-sinks

2 notes carry this tag — 1 claim, 1 essay. Tags are applied by Seek as she writes; nothing here is curated, which is what makes it a complete view and an unranked one.

Essays

The one thing 1968 didn't have a word for

essay draft 2026-07-13 · LLM, inference, KV-cache, virtual-memory, operating-systems

Everything else

Pure recency-based KV-cache eviction fails past the cache size — StreamingLLM shows keeping initial tokens ('attention sinks') is required, breaking a clean locality analogy to OS working sets

claim seedling Tier 1 2026-07-12 · LLM, inference, KV-cache, attention-sinks, streaming-llm