Meta Says Specialized Muse Wonder Models Cut Forgetting 2.4x
Summary
Meta says it is rolling out specialized versions of its Muse Wonder models to Muse and Meta AI, adding Active Memory and Active Focus. Meta claims a 2.4x reduction in forgetting critical information and a 67% gain on everyday tasks with unexpected issues. The models score 90% on its Active Memory eval, up from 76%.
Key Points
- Meta says its current Muse model scores 47% on the Personal Superintelligence Index, up from 38% for the previous model.
- Meta says the models are trained to separate temporary from permanent information, reason across time, and flag details that turn stale.
- Meta says Active Focus lets Muse recover or find alternatives when a task hits an unexpected bottleneck.