config.json

{
  "architectures": [
    "SamHQVisionModel"
  ],
  "attention_dropout": 0.0,
  "dropout": 0.0,
  "dtype": "float32",
  "global_attn_indexes": [
    2,
    5,
    8,
    11
  ],
  "hidden_act": "gelu",
  "hidden_size": 36,
  "image_size": 24,
  "initializer_factor": 1.0,
  "initializer_range": 0.02,
  "intermediate_size": 72,
  "layer_norm_eps": 1e-06,
  "mlp_dim": 144,
  "mlp_ratio": 4.0,
  "model_type": "sam_hq_vision_model",
  "num_attention_heads": 4,
  "num_channels": 3,
  "num_hidden_layers": 12,
  "num_pos_feats": 16,
  "output_channels": 32,
  "patch_size": 2,
  "projection_dim": 62,
  "qkv_bias": true,
  "rel_pos_zero_init": false,
  "scale": 18,
  "transformers_version": "5.16.0.dev0",
  "use_abs_pos": true,
  "use_rel_pos": true,
  "window_size": 14
}