Nolege News

Tag

#byte-pair encoding

Every article tagged “byte-pair encoding”.

1 article

Ask a chatbot how many r's are in \

Computer Science

Ask a chatbot how many r's are in \"strawberry\". It struggles because it has never seen a letter

20 Sept 2026 · 8 min read

A model that can discuss differential equations will miscount the letters in a common word. That is not a gap in its reasoning — it is a consequence of the vocabulary that was fixed before training began, and the same choice makes Hindi several times more expensive to process than English.

Read more →

Get the science, minus the hype

Clear, occasional updates on nanotech, materials and research. No spam, unsubscribe anytime.