config.json

{
  "apply_layernorm_after_vision_backbone": true,
  "architectures": [
    "OmDetTurboForObjectDetection"
  ],
  "backbone": "swin_tiny_patch4_window7_224",
  "backbone_config": null,
  "backbone_depths": [
    2,
    2,
    6,
    2
  ],
  "backbone_embed_dim": 96,
  "backbone_image_size": 640,
  "backbone_kwargs": {
    "always_partition": true,
    "img_size": 640,
    "out_indices": [
      1,
      2,
      3
    ]
  },
  "backbone_num_heads": [
    3,
    6,
    12,
    24
  ],
  "backbone_out_indices": [
    1,
    2,
    3
  ],
  "backbone_window_size": 7,
  "batch_norm_eps": 1e-05,
  "cache_size": 100,
  "class_distance_type": "cosine",
  "class_embed_dim": 512,
  "conv_norm_activation": "gelu",
  "csp_activation": "silu",
  "d_model": 256,
  "decoder_activation": "relu",
  "decoder_dim_feedforward": 2048,
  "decoder_dropout": 0.0,
  "decoder_hidden_dim": 256,
  "decoder_num_heads": 8,
  "decoder_num_layers": 6,
  "decoder_num_points": 4,
  "disable_custom_kernels": false,
  "encoder_attention_heads": 8,
  "encoder_dim_feedforward": 2048,
  "encoder_dropout": 0.0,
  "encoder_feedforward_activation": "relu",
  "encoder_feedforward_dropout": 0.0,
  "encoder_in_channels": [
    192,
    384,
    768
  ],
  "encoder_layers": 1,
  "encoder_projection_indices": [
    2
  ],
  "eval_size": null,
  "hidden_expansion": 1,
  "image_size": 640,
  "init_std": 0.02,
  "is_encoder_decoder": true,
  "layer_norm_eps": 1e-05,
  "learn_initial_query": false,
  "model_type": "omdet-turbo",
  "num_feature_levels": 3,
  "num_queries": 900,
  "positional_encoding_temperature": 10000,
  "task_encoder_hidden_dim": 1024,
  "text_config": {
    "model_type": "clip_text_model"
  },
  "text_projection_in_dim": 512,
  "text_projection_out_dim": 512,
  "torch_dtype": "float32",
  "transformers_version": "4.45.0.dev0",
  "use_pretrained_backbone": false,
  "use_timm_backbone": true,
  "vision_features_channels": [
    256,
    256,
    256
  ]
}