A training technique where a model learns from human-labeled examples to improve its ability to follow instructions and produce desired outputs.
Adhering to complex, structured, or constrained instructions