{
    "architectures": [
        "YolosForObjectDetection"
    ],
    "model_type": "yolos",
    "attention_probs_dropout_prob": 0.1,
    "hidden_dropout_prob": 0.1,
    "hidden_size": 384,
    "image_size": [
        640,
        640
    ],
    "initializer_range": 0.02,
    "layer_norm_eps": 1e-12,
    "num_attention_heads": 6,
    "num_channels": 3,
    "num_hidden_layers": 12,
    "patch_size": 16,
    "qkv_bias": true,
    "encoder_stride": 16,
    "num_detection_tokens": 100,
    "intermediate_size": 1536,
    "id2label": {
        "0": "roof"
    },
    "label2id": {
        "roof": 0
    },
    "use_mid_position_embeddings": true,
    "use_relative_position_bias": true,
    "bbox_cost": 5,
    "bbox_loss_coefficient": 5,
    "class_cost": 1,
    "giou_cost": 2,
    "giou_loss_coefficient": 2
}