other8.0 / 10
Safety Pretraining
Abstract
A novel set of interventions during data curation and pretraining to instill safer model behavior from the start
Research area
agentic misalignmentpost-trainingpretraining
Published
—
Source
other
Org
Gray Swan AI
Sign in to read and join the discussion.