AINewsnow

Catastrophic Forgetting, Measured: I Fine-Tuned a Small LLM on One New Character and Its Old Stories Went From Perplexity 9 to 156

Part 4 of ten. Each post is a short, plain-words tour of one lesson from e=mc² , a free build-an-LLM-from-scratch course; the full lesson, notebook and numbers are on the site. The most common thing anyone does with a language model is teach it something new, and almost nobody measures what that co…

Read the full story at DEV Community — Machine Learning ↗

Timeline · 1 report

  1. 2026-11-18 12:30 · DEV Community — Machine Learning
    Catastrophic Forgetting, Measured: I Fine-Tuned a Small LLM on One New Character and Its Old Stories Went From Perplexity 9 to 156

More stories

  1. Study of 300 Product-Fact Queries Finds AI Engines Often Agree Without Being Right — Unite.AI
  2. Perplexity Launches Orchestrator API With Same-Model Failover — Unite.AI
  3. Perplexity CEO hints at possible 2027 IPO as annual revenue crosses $1 billion — CNBC Technology
  4. How AI Answer Engines Read Your WordPress Site (and How to Make Them Read It Right) — DEV Community — AI
  5. Perplexity Publishes Guide to Managed RAG for Enterprise AI — Unite.AI
  6. Is that really GPTBot? How to see which AI agents visit your site and which ones are fake — DEV Community — AI
  7. I checked 10 SEO agency websites for AI-search readiness. The real bugs were the boring ones. — DEV Community — AI
  8. Perplexity Adds Voice Control for AI Teammates on iOS and Android — Unite.AI

Read the Daily Brief: the day's most important AI stories →