{
  "library_name": "kerasformers",
  "kerasformers_version": "1.2.1",
  "model_module": "kerasformers.models.grounding_dino",
  "model_class": "GroundingDinoDetect",
  "variant": "grounding_dino_tiny",
  "weights": "model.weights.h5",
  "schema_version": 2,
  "weight_dtype": "float32",
  "model_type": "grounding-dino",
  "vision_config": {
    "d_model": 256,
    "encoder_layers": 6,
    "encoder_ffn_dim": 2048,
    "encoder_attention_heads": 8,
    "decoder_layers": 6,
    "decoder_ffn_dim": 2048,
    "decoder_attention_heads": 8,
    "num_queries": 900,
    "num_feature_levels": 4,
    "encoder_n_points": 4,
    "decoder_n_points": 4,
    "max_text_len": 256,
    "query_dim": 4,
    "two_stage": true,
    "positional_embedding_temperature": 20.0,
    "layer_norm_eps": 1e-05,
    "activation_function": "relu",
    "backbone_embed_dim": 96,
    "backbone_depths": [
      2,
      2,
      6,
      2
    ],
    "backbone_num_heads": [
      3,
      6,
      12,
      24
    ],
    "backbone_window_size": 7,
    "backbone_out_indices": [
      2,
      3,
      4
    ],
    "text_vocab_size": 30522,
    "text_hidden_size": 768,
    "text_num_layers": 12,
    "text_num_heads": 12,
    "text_intermediate_size": 3072,
    "text_max_position_embeddings": 512,
    "text_layer_norm_eps": 1e-12
  }
}