Skip to content
Research

Modern LLMs have tiny GPTs hidden inside them.

LessWrongAnalysis or commentary··Updated yesterday
AI brief

The post describes experiments using Qwen models to predict GPT2 completions, and says it is a crosspost from a substack and part of Lossfunk, where meta-cognition in LLMs is being investigated.

Written by AI from LessWrong's published text. Read the original for full details.

Source

Interpretation or community commentary rather than straight reporting.

Read original story ↗