MultimodalTransformer constructor

MultimodalTransformer({
  1. required AudioTransformer audioModel,
  2. required VideoTransformer videoModel,
  3. required TextTransformer textModel,
  4. required int numClasses,
})

Implementation

MultimodalTransformer({
  required this.audioModel,
  required this.videoModel,
  required this.textModel,
  required int numClasses,
}) : fusionLayer = Layer(
       // Logic: Audio(64) + Video(128) + Text(128) = 320
       audioModel.embedSize + videoModel.embedSize + textModel.embedSize,
       numClasses,
       useGelu: false,
     );