config.json 2.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130
  1. {
  2. "activation_dropout": 0.0,
  3. "activation_function": "silu",
  4. "anchor_image_size": null,
  5. "architectures": [
  6. "RTDetrForObjectDetection"
  7. ],
  8. "attention_dropout": 0.0,
  9. "auxiliary_loss": true,
  10. "backbone": null,
  11. "backbone_config": {
  12. "model_type": "rt_detr_resnet",
  13. "out_features": [
  14. "stage2",
  15. "stage3",
  16. "stage4"
  17. ],
  18. "out_indices": [
  19. 2,
  20. 3,
  21. 4
  22. ]
  23. },
  24. "backbone_kwargs": null,
  25. "batch_norm_eps": 1e-05,
  26. "box_noise_scale": 1.0,
  27. "d_model": 256,
  28. "decoder_activation_function": "relu",
  29. "decoder_attention_heads": 8,
  30. "decoder_ffn_dim": 1024,
  31. "decoder_in_channels": [
  32. 256,
  33. 256,
  34. 256
  35. ],
  36. "decoder_layers": 6,
  37. "decoder_n_points": 4,
  38. "disable_custom_kernels": true,
  39. "dropout": 0.0,
  40. "encode_proj_layers": [
  41. 2
  42. ],
  43. "encoder_activation_function": "gelu",
  44. "encoder_attention_heads": 8,
  45. "encoder_ffn_dim": 1024,
  46. "encoder_hidden_dim": 256,
  47. "encoder_in_channels": [
  48. 512,
  49. 1024,
  50. 2048
  51. ],
  52. "encoder_layers": 1,
  53. "eos_coefficient": 0.0001,
  54. "eval_size": null,
  55. "feat_strides": [
  56. 8,
  57. 16,
  58. 32
  59. ],
  60. "focal_loss_alpha": 0.75,
  61. "focal_loss_gamma": 2.0,
  62. "freeze_backbone_batch_norms": true,
  63. "hidden_expansion": 1.0,
  64. "id2label": {
  65. "0": "background",
  66. "1": "Caption",
  67. "10": "Text",
  68. "11": "Title",
  69. "12": "Document Index",
  70. "13": "Code",
  71. "14": "Checkbox-Selected",
  72. "15": "Checkbox-Unselected",
  73. "16": "Form",
  74. "17": "Key-Value Region",
  75. "2": "Footnote",
  76. "3": "Formula",
  77. "4": "List-item",
  78. "5": "Page-footer",
  79. "6": "Page-header",
  80. "7": "Picture",
  81. "8": "Section-header",
  82. "9": "Table"
  83. },
  84. "initializer_bias_prior_prob": null,
  85. "initializer_range": 0.01,
  86. "is_encoder_decoder": true,
  87. "label2id": {
  88. "Caption": "1",
  89. "Checkbox-Selected": "14",
  90. "Checkbox-Unselected": "15",
  91. "Code": "13",
  92. "Document Index": "12",
  93. "Footnote": "2",
  94. "Form": "16",
  95. "Formula": "3",
  96. "Key-Value Region": "17",
  97. "List-item": "4",
  98. "Page-footer": "5",
  99. "Page-header": "6",
  100. "Picture": "7",
  101. "Section-header": "8",
  102. "Table": "9",
  103. "Text": "10",
  104. "Title": "11",
  105. "background": "0"
  106. },
  107. "label_noise_ratio": 0.5,
  108. "layer_norm_eps": 1e-05,
  109. "learn_initial_query": false,
  110. "matcher_alpha": 0.25,
  111. "matcher_bbox_cost": 5.0,
  112. "matcher_class_cost": 2.0,
  113. "matcher_gamma": 2.0,
  114. "matcher_giou_cost": 2.0,
  115. "model_type": "rt_detr",
  116. "normalize_before": false,
  117. "num_denoising": 100,
  118. "num_feature_levels": 3,
  119. "num_queries": 300,
  120. "positional_encoding_temperature": 10000,
  121. "torch_dtype": "float32",
  122. "transformers_version": "4.46.2",
  123. "use_focal_loss": true,
  124. "use_pretrained_backbone": false,
  125. "use_timm_backbone": false,
  126. "weight_loss_bbox": 5.0,
  127. "weight_loss_giou": 2.0,
  128. "weight_loss_vfl": 1.0,
  129. "with_box_refine": true
  130. }