{
  "models": {
    "small": {
      "patch_size": 16,
      "embed_dim": 512,
      "depth": 8,
      "num_heads": 8,
      "mlp_ratio": 4.0,
      "decoder_embed_dim": 384,
      "decoder_depth": 4,
      "decoder_num_heads": 16
    },
    "base": {
      "patch_size": 16,
      "embed_dim": 768,
      "depth": 12,
      "num_heads": 12,
      "mlp_ratio": 4.0,
      "decoder_embed_dim": 512,
      "decoder_depth": 8,
      "decoder_num_heads": 16
    },
    "large": {
      "patch_size": 16,
      "embed_dim": 1024,
      "depth": 24,
      "num_heads": 16,
      "mlp_ratio": 4.0,
      "decoder_embed_dim": 512,
      "decoder_depth": 8,
      "decoder_num_heads": 16
    },
    "largeV2": {
      "patch_size": 16,
      "embed_dim": 1024,
      "depth": 24,
      "num_heads": 16,
      "mlp_ratio": 4.0,
      "decoder_embed_dim": 512,
      "decoder_depth": 8,
      "decoder_num_heads": 16
    }
  }
}