Meta Brings Muse Wonder Memory and Focus to Muse and Meta AI
Source
research.meta.ai
Author
AIatMeta
Date
Why it matters
Meta ships models trained for AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → memory and recovery from blocked tasks, with self-reported gains (2.4x less forgetting of critical context windowThe maximum amount of text a model can consider at once — its working memory for the current conversation or task.Full definition →, 67% better recovery). It shows how labs target long-running agent reliability.
Terms in this piece · Glossary
eval — A repeatable test for AI quality — a set of tasks plus scoring — used the way software teams use test suites, because model output is too variable to judge by eyeballing.
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
