##article.return## Momentum-Guided Semantic Forecasting (MoFore) for Self-Supervised Video Representation Learning Download Download PDF