scikit-llm

Why we included this project

Data scientists who already work in scikit-learn can add LLM-powered text analysis without learning a new framework. The library wraps models like GPT and Vertex AI in familiar estimator classes, so a ZeroShotGPTClassifier behaves like any other scikit-learn model: you call fit() and predict() the same way you always have. That covers zero-shot and few-shot classification, multi-label tagging, and text labeling, all inside your existing pipelines. It also runs against locally hosted backends, which is useful when data has to stay on-premises or you want to avoid per-token API costs. For teams that want to experiment with LLMs but stay on a proven API surface, this is a practical bridge rather than a new stack to adopt.