Zero-Shot Rationale Distillation Cuts Model Annotation Costs While Preserving Task Accuracy
Researchers demonstrate that distilling large language models using automated zero-shot reasoning rationales reduces annotation token expenses tenfold while matching the ...

