Adaptive Classifier is a machine learning framework for dynamic text classification built on PyTorch and HuggingFace transformers. It supports continuous and online learning with dynamic class addition and employs elastic weight consolidation for catastrophic forgetting defense. The library includes ONNX Runtime optimization for faster inference and integrates FAISS for embedding management, compatible with any HuggingFace transformer model.