Data Repetition Beats Data Scaling in Long-CoT Supervised Fine-Tuning
IntermediateDawid J. Kopiczko, Sagar Vaze et al.Feb 11arXiv
The paper shows that, when teaching a reasoning AI with step-by-step examples, repeating a small set many times can beat using a huge set only once.
#Supervised Fine-Tuning#Chain-of-Thought#Data Repetition