Local Minimum

Lowest possible loss.

Flat high-dimensional regions slowing training.

Choosing step size along gradient direction.

Restricting updates to safe regions.

Forcing predictable formats for downstream systems; reduces parsing errors and supports validation/guardrails.

Models that process or generate multiple modalities, enabling vision-language tasks, speech, video understanding, etc.

AI subfield dealing with understanding and generating human language, including syntax, semantics, and pragmatics.

Visualization of optimization landscape.