{
  "architectures": [
    "MobileViTV2ForSemanticSegmentation"
  ],
  "aspp_dropout_prob": 0.1,
  "aspp_out_channels": 32,
  "atrous_rates": [
    6,
    12,
    18
  ],
  "attn_dropout": 0.0,
  "base_attn_unit_dims": [
    16,
    24,
    32
  ],
  "classifier_dropout_prob": 0.1,
  "conv_kernel_size": 3,
  "dtype": "float32",
  "expand_ratio": 2.0,
  "ffn_dropout": 0.0,
  "ffn_multiplier": 2,
  "hidden_act": "swish",
  "image_size": 64,
  "initializer_range": 0.02,
  "layer_norm_eps": 1e-05,
  "model_type": "mobilevitv2",
  "n_attn_blocks": [
    1,
    1,
    2
  ],
  "num_channels": 3,
  "output_stride": 32,
  "patch_size": 2,
  "semantic_loss_ignore_index": 255,
  "transformers_version": "5.16.0.dev0",
  "width_multiplier": 0.25
}