cd/entity/SRT-Introspectยท homeโ€บ entitiesโ€บ SRT-Introspect
grep -l @srt-introspect /news/*.json | wc -l โ†’ 1

@SRT-Introspect

mentions 1 type Organization feed RSS
19:33
2026-06-04
news.ycombinator.com
large-language-models

Words do not have determined meanings

A new demo called SRT-Introspect reveals how large language models like Qwen fix word meanings during generation, contradicting the reflexive, approximated nature of human language. The tool surfaces โ€ฆ

// co-occurs with top 3 entities