config.json

{
  "activation_dropout": 0.0,
  "activation_function": "silu",
  "anchor_image_size": null,
  "architectures": [
    "PPDocLayoutV3ForObjectDetection"
  ],
  "attention_dropout": 0.0,
  "backbone_config": {
    "arch": "L",
    "depths": [
      3,
      4,
      6,
      3
    ],
    "embedding_size": 64,
    "freeze_at": 0,
    "freeze_norm": true,
    "freeze_stem_only": true,
    "hidden_act": "relu",
    "hidden_sizes": [
      32,
      32,
      32,
      32
    ],
    "initializer_range": 0.02,
    "lr_mult_list": [
      0,
      0.05,
      0.05,
      0.05,
      0.05
    ],
    "model_type": "hgnet_v2",
    "num_channels": 3,
    "out_features": [
      "stage1",
      "stage2",
      "stage3",
      "stage4"
    ],
    "out_indices": [
      1,
      2,
      3,
      4
    ],
    "return_idx": [
      0,
      1,
      2,
      3
    ],
    "stage_downsample": [
      false,
      true,
      true,
      true
    ],
    "stage_downsample_strides": [
      2,
      2,
      2,
      2
    ],
    "stage_in_channels": [
      32,
      32,
      32,
      32
    ],
    "stage_kernel_size": [
      3,
      3,
      5,
      5
    ],
    "stage_light_block": [
      false,
      false,
      true,
      true
    ],
    "stage_mid_channels": [
      32,
      32,
      32,
      32
    ],
    "stage_names": [
      "stem",
      "stage1",
      "stage2",
      "stage3",
      "stage4"
    ],
    "stage_num_blocks": [
      1,
      1,
      3,
      1
    ],
    "stage_numb_of_layers": [
      6,
      6,
      6,
      6
    ],
    "stage_out_channels": [
      32,
      32,
      32,
      32
    ],
    "stem_channels": [
      3,
      32,
      32
    ],
    "stem_strides": [
      2,
      1,
      1,
      2,
      1
    ],
    "use_learnable_affine_block": false
  },
  "batch_norm_eps": 1e-05,
  "box_noise_scale": 1.0,
  "d_model": 32,
  "decoder_activation_function": "relu",
  "decoder_attention_heads": 2,
  "decoder_ffn_dim": 8,
  "decoder_in_channels": [
    32,
    32,
    32
  ],
  "decoder_layers": 2,
  "decoder_n_points": 4,
  "disable_custom_kernels": true,
  "dropout": 0.0,
  "dtype": "float32",
  "encode_proj_layers": [
    2
  ],
  "encoder_activation_function": "gelu",
  "encoder_attention_heads": 2,
  "encoder_ffn_dim": 64,
  "encoder_hidden_dim": 32,
  "encoder_in_channels": [
    20,
    30,
    40
  ],
  "encoder_layers": 1,
  "eval_size": null,
  "feat_strides": [
    8,
    16,
    32
  ],
  "freeze_backbone_batch_norms": true,
  "global_pointer_head_size": 64,
  "gp_dropout_value": 0.1,
  "hidden_expansion": 1.0,
  "image_size": 128,
  "initializer_bias_prior_prob": null,
  "initializer_range": 0.01,
  "is_encoder_decoder": true,
  "label_noise_ratio": 0.5,
  "layer_norm_eps": 1e-05,
  "learn_initial_query": false,
  "mask_enhanced": true,
  "mask_feature_channels": [
    32,
    32
  ],
  "model_type": "pp_doclayout_v3",
  "normalize_before": false,
  "num_denoising": 0,
  "num_feature_levels": 3,
  "num_prototypes": 32,
  "num_queries": 30,
  "positional_encoding_temperature": 10000,
  "tie_word_embeddings": true,
  "transformers_version": "5.16.0.dev0",
  "x4_feat_dim": 32
}