config.json

{
  "architectures": [
    "Sam2VisionModel"
  ],
  "backbone_channel_list": [
    96,
    48,
    24,
    12
  ],
  "backbone_config": {
    "blocks_per_stage": [
      1,
      2,
      7,
      2
    ],
    "embed_dim_per_stage": [
      12,
      24,
      48,
      96
    ],
    "global_attention_blocks": [
      5,
      7,
      9
    ],
    "hidden_act": "gelu",
    "hidden_size": 12,
    "image_size": 128,
    "initializer_range": 0.02,
    "layer_norm_eps": 1e-06,
    "mlp_ratio": 4.0,
    "model_type": "sam2_hiera_det_model",
    "num_attention_heads": 1,
    "num_attention_heads_per_stage": [
      1,
      2,
      4,
      8
    ],
    "num_channels": 3,
    "num_query_pool_stages": 3,
    "patch_kernel_size": 7,
    "patch_padding": 3,
    "patch_stride": 4,
    "query_stride": [
      2,
      2
    ],
    "window_positional_embedding_background_size": [
      7,
      7
    ],
    "window_size_per_stage": [
      8,
      4,
      14,
      7
    ]
  },
  "backbone_feature_sizes": [
    [
      32,
      32
    ],
    [
      16,
      16
    ],
    [
      8,
      8
    ]
  ],
  "dtype": "float32",
  "fpn_hidden_size": 32,
  "fpn_kernel_size": 1,
  "fpn_padding": 0,
  "fpn_stride": 1,
  "fpn_top_down_levels": [
    2,
    3
  ],
  "hidden_act": "gelu",
  "initializer_range": 0.02,
  "layer_norm_eps": 1e-06,
  "model_type": "sam2_vision_model",
  "num_feature_levels": 3,
  "transformers_version": "5.16.0.dev0"
}