Named Entity Recognition (NER) is a crucial task in natural language processing that involves identifying and classifying entities such as names, places, and dates within text. While Large Language Models (LLMs) have recently gained attention for their effectiveness in NER, simpler and cost-effective methods have existed for some time. Few-shot NER, in particular, addresses the challenge of adapting models to new, domain-specific entity types with minimal training data. Techniques such as prototypical networks, contrastive learning, and meta-learning have been explored to overcome the limitations of traditional deep learning approaches, which typically require large datasets. Recent advancements in LLMs show promise for few-shot NER, as they excel at learning from limited examples, although they face challenges due to the token-level nature of NER tasks. The Few-NERD dataset has become a popular benchmark for evaluating few-shot NER systems, highlighting the field's progress and the potential for future developments in making these systems more adaptable and efficient.