AINewsnow

Why Can't LLMs Count the R's in "Strawberry"? How Tokenizers Work

The preprocessing step behind spelling slips, shaky arithmetic, glitch tokens and your API bill Ask a language model how many r's are in "strawberry" and, for a long time, the popular answer was "two." The same models could write a sonnet about strawberries, explain their genetics and translate the…

Read the full story at DEV Community — Machine Learning ↗

Timeline · 1 report

  1. 2026-10-13 22:49 · DEV Community — Machine Learning
    Why Can't LLMs Count the R's in "Strawberry"? How Tokenizers Work

More stories

  1. Google Rolls Out Gemini 4 Argon Frontier Model — Google AI Blog
  2. Apple launches HomeView, Apple TV 4K, and HomePod mini 2 with Siri AI — The Guardian AI
  3. FBI arrests OpenAI employee for planting recording device in office — Sky News Technology
  4. Salesforce launches 10 new Service Cloud AI capabilities — Salesforce AI Blog
  5. Here’s how to find the learning strategy that makes AI work for you — The Conversation AI (US)
  6. Developing Gemini 4 Argon responsibly and safely — Google Gemini Blog
  7. Google Play enables private installation of AI-generated Android apps — TechCrunch AI
  8. Anthropic warns AI models could self-replicate across smart devices — CNBC Technology

Get the daily brief of stories like this at 6:30 every morning →