config.json

{
  "architectures": [
    "MaskFormerModel"
  ],
  "backbone": null,
  "backbone_config": {
    "attention_probs_dropout_prob": 0.0,
    "depths": [
      1,
      1,
      1,
      1
    ],
    "drop_path_rate": 0.1,
    "embed_dim": 16,
    "encoder_stride": 32,
    "hidden_act": "gelu",
    "hidden_dropout_prob": 0.0,
    "hidden_size": 32,
    "image_size": 224,
    "initializer_range": 0.02,
    "layer_norm_eps": 1e-05,
    "mlp_ratio": 4.0,
    "model_type": "maskformer-swin",
    "num_channels": 3,
    "num_heads": [
      1,
      1,
      2,
      2
    ],
    "num_layers": 4,
    "out_features": [
      "stage1",
      "stage2",
      "stage3",
      "stage4"
    ],
    "out_indices": [
      1,
      2,
      3,
      4
    ],
    "patch_size": 4,
    "qkv_bias": true,
    "stage_names": [
      "stem",
      "stage1",
      "stage2",
      "stage3",
      "stage4"
    ],
    "use_absolute_embeddings": false,
    "window_size": 7
  },
  "cross_entropy_weight": 1.0,
  "decoder_config": {
    "activation_dropout": 0.0,
    "activation_function": "relu",
    "attention_dropout": 0.0,
    "auxiliary_loss": false,
    "backbone_config": {
      "_name_or_path": "",
      "architectures": null,
      "backbone": "resnet50",
      "chunk_size_feed_forward": 0,
      "dtype": null,
      "features_only": true,
      "freeze_batch_norm_2d": false,
      "id2label": {
        "0": "LABEL_0",
        "1": "LABEL_1"
      },
      "is_encoder_decoder": false,
      "label2id": {
        "LABEL_0": 0,
        "LABEL_1": 1
      },
      "model_type": "timm_backbone",
      "num_channels": 3,
      "out_features": null,
      "out_indices": [
        1,
        2,
        3,
        4
      ],
      "output_attentions": false,
      "output_hidden_states": false,
      "output_stride": null,
      "problem_type": null,
      "return_dict": true,
      "use_pretrained_backbone": false
    },
    "bbox_cost": 5,
    "bbox_loss_coefficient": 5,
    "class_cost": 1,
    "d_model": 32,
    "decoder_attention_heads": 2,
    "decoder_ffn_dim": 64,
    "decoder_layerdrop": 0.0,
    "decoder_layers": 2,
    "dice_loss_coefficient": 1,
    "dilation": false,
    "dropout": 0.1,
    "encoder_attention_heads": 2,
    "encoder_ffn_dim": 64,
    "encoder_layerdrop": 0.0,
    "encoder_layers": 2,
    "eos_coefficient": 0.1,
    "giou_cost": 2,
    "giou_loss_coefficient": 2,
    "init_std": 0.02,
    "init_xavier_std": 1.0,
    "mask_loss_coefficient": 1,
    "model_type": "detr",
    "num_channels": 3,
    "num_queries": 10,
    "position_embedding_type": "sine"
  },
  "dice_weight": 1.0,
  "dtype": "float32",
  "fpn_feature_size": 32,
  "id2label": {
    "0": "LABEL_0",
    "1": "LABEL_1",
    "2": "LABEL_2",
    "3": "LABEL_3"
  },
  "init_std": 0.02,
  "init_xavier_std": 1.0,
  "label2id": {
    "LABEL_0": 0,
    "LABEL_1": 1,
    "LABEL_2": 2,
    "LABEL_3": 3
  },
  "mask_feature_size": 32,
  "mask_weight": 20.0,
  "model_type": "maskformer",
  "no_object_weight": 0.1,
  "num_attention_heads": 2,
  "num_channels": 3,
  "num_hidden_layers": 2,
  "output_auxiliary_logits": null,
  "transformers_version": "5.16.0.dev0",
  "use_auxiliary_loss": false
}