This repository provides a standardized benchmark for evaluating semi-supervised learning methods across multiple modalities including computer vision, natural language processing, and audio classification. It unifies various SSL algorithms and evaluation protocols to enable fair comparison. The project is implemented in PyTorch and supports modern transformer architectures.