config.json

{
  "architectures": [
    "Data2VecVisionModel"
  ],
  "attention_probs_dropout_prob": 0.0,
  "auxiliary_channels": 256,
  "auxiliary_concat_input": false,
  "auxiliary_loss_weight": 0.4,
  "auxiliary_num_convs": 1,
  "drop_path_rate": 0.1,
  "hidden_act": "gelu",
  "hidden_dropout_prob": 0.0,
  "hidden_size": 1024,
  "image_size": 224,
  "initializer_range": 0.02,
  "intermediate_size": 4096,
  "layer_norm_eps": 1e-12,
  "layer_scale_init_value": 0.1,
  "model_type": "data2vec-vision",
  "num_attention_heads": 16,
  "num_channels": 3,
  "num_hidden_layers": 24,
  "out_indices": [
    3,
    5,
    7,
    11
  ],
  "patch_size": 16,
  "pool_scales": [
    1,
    2,
    3,
    6
  ],
  "semantic_loss_ignore_index": 255,
  "torch_dtype": "float32",
  "transformers_version": "4.19.0.dev0",
  "use_absolute_position_embeddings": false,
  "use_auxiliary_head": true,
  "use_mask_token": false,
  "use_mean_pooling": false,
  "use_relative_position_bias": false,
  "use_shared_relative_position_bias": true,
  "vocab_size": 8192
}