Machine Learning & Training

What Is Self-Supervised Learning?

Self-supervised learning is a technique where models generate their own training signals from unlabeled data. By solving tasks such as predicting missing parts of the input, models learn useful representations without manual labels. This approach underpins the pretraining of many modern language and vision models.