Welcome to USB’s documentation!

Unified Semi-supervised learning Benchmark (USB) is a modular and extensible codebase including data pipeline, and popular SSL algorithms for standardization of SSL ablations. Meanwhile, pre-trained versions of the state-of-the-art neural models for CV tasks are provided. It is easy-to-use/extend, affordable, and comprehensive for developing and evaluating SSL algorithms. USB provides the implementation of 14 SSL algorithms based on Consistency Regularization, and 15 tasks for evaluation from CV, NLP, and Audio domain.

We provide a Python package semilearn of USB for users who want to start training/testing the supported SSL algorithms on their data quickly:

` pip install semilearn `

For detailed information on the various tasks over different domains, the developed algorithms and the customized usage, please refer to the following sections.

• Datasets Zoo
  • CV Datasets
  • NLP Datasets
  • Audio Datasets
• Algorithm Zoo
• Tutorial
  • Get Started
  • Customized Usage
• API
  • Algorithms
  • Datasets
  • Model Zoo
  • Lighting