configs/model/info_sharing/aat_ifr_24_layers.yaml

# Model type (Options: ["cross_attention", "global_attention", "alternating_attention"])
model_type: "alternating_attention"
# Model class type (Options: ["no_intermediate_features", "intermediate_features"])
model_return_type: "intermediate_features"
# Custom positional encoding (Options: ["RoPEfreq"], Callable Function, null)
custom_positional_encoding: null
# Module arguments
module_args:
  # Name of the info sharing module
  name: "aat_24_layers_ifr"
  # Indices of the intermediate features to be shared (indices start from 0)
  indices: [11, 17]
  # Normalize intermediate features
  norm_intermediate: True
  # Size string
  size: "24_layers"
  # Depth
  depth: 24
  # Distinguish Reference and Non-Reference Views
  distinguish_ref_and_non_ref_views: True
  # Flag to indicate whether to use gradient checkpointing
  gradient_checkpointing: False