LibFewShot provides a unified framework for few-shot learning research, implementing methods across three paradigms: non-episodic fine-tuning approaches, meta-learning algorithms like MAML and MatchingNet, and metric-learning methods such as ProtoNet and RelationNet. The library standardizes benchmarking across standard few-shot datasets (miniImageNet, TieredImageNet, CUB) and is published in IEEE TPAMI 2023.