STAA: Spatio-Temporal Attention Attribution for Real-Time Interpreting Transformer-based Video Models

Open in new window