{
  "architectures": [
    "EfficientFormerForImageClassificationWithTeacher"
  ],
  "attention_probs_dropout_prob": 0.1,
  "attention_ratio": 4,
  "depths": [
    3,
    2,
    6,
    4
  ],
  "dim": 448,
  "distillation": true,
  "downsample_pad": 1,
  "downsample_patch_size": 3,
  "downsample_stride": 2,
  "downsamples": [
    true,
    true,
    true,
    true
  ],
  "drop_path_rate": 0.0,
  "encoder_stride": 2,
  "hidden_act": "gelu",
  "hidden_dropout_prob": 0.1,
  "hidden_size": 448,
  "hidden_sizes": [
    48,
    96,
    224,
    448
  ],
  "image_size": 224,
  "initializer_range": 0.02,
  "intermediate_size": 37,
  "key_dim": 32,
  "layer_norm_eps": 1e-12,
  "layer_scale_init_value": 1e-05,
  "mlp_expansion_ratio": 4,
  "model_type": "efficientformer",
  "num_attention_heads": 8,
  "num_channels": 3,
  "num_hidden_layers": 7,
  "num_meta3d_blocks": 1,
  "patch_size": 2,
  "pool_size": 3,
  "resolution": 7,
  "torch_dtype": "float32",
  "transformers_version": "4.28.0.dev0",
  "use_layer_scale": true
}