Prefix-RFT: A Unified Machine Learning Framework to blend Supervised Fine-Tuning (SFT) and Reinforcement Fine-Tuning (RFT)

Large language models are typically refined after pretraining using either supervised fine-tuning (SFT) or reinforcement fine-tuning (RFT), each with distinct strengths and limitations. SFT is effective in teaching instruction-following through example-based learning, but it can lead to rigid behavior and poor generalization. RFT, on the other hand, optimizes models for task success using reward signals, […]

The post Prefix-RFT: A Unified Machine Learning Framework to blend Supervised Fine-Tuning (SFT) and Reinforcement Fine-Tuning (RFT) appeared first on MarkTechPost.