fairseq/examples/MMPT/projects/task/coin_videoclip.yaml

includes: projects/task/coin.yaml
model:
  model_cls: MMFusionSeparateActionSegmentation
  mm_encoder_cls: 
  video_encoder_cls: MMBertForTokenClassification
  text_encoder_cls: BertModel  # dummy, not used.
  num_hidden_video_layers: 6