config.json

{
  "architectures": [
    "Sam3VisionModel"
  ],
  "backbone_config": {
    "attention_dropout": 0.0,
    "global_attn_indexes": [
      0,
      1
    ],
    "hidden_act": "gelu",
    "hidden_dropout": 0.0,
    "hidden_size": 32,
    "image_size": 224,
    "initializer_range": 0.02,
    "intermediate_size": 64,
    "layer_norm_eps": 1e-06,
    "layer_scale_init_value": null,
    "model_type": "sam3_vit_model",
    "num_attention_heads": 4,
    "num_channels": 3,
    "num_hidden_layers": 2,
    "patch_size": 14,
    "pretrain_image_size": 336,
    "rope_theta": 10000.0,
    "window_size": 8
  },
  "backbone_feature_sizes": [
    [
      288,
      288
    ],
    [
      144,
      144
    ],
    [
      72,
      72
    ]
  ],
  "dtype": "float32",
  "fpn_hidden_size": 32,
  "hidden_act": "gelu",
  "initializer_range": 0.02,
  "layer_norm_eps": 1e-06,
  "model_type": "sam3_vision_model",
  "scale_factors": [
    2.0,
    1.0
  ],
  "transformers_version": "5.16.0.dev0"
}