config.json

{
  "_name_or_path": "microsoft/layoutxlm-base",
  "architectures": [
    "LayoutLMv2ForTokenClassification"
  ],
  "attention_probs_dropout_prob": 0.1,
  "bos_token_id": 0,
  "convert_sync_batchnorm": true,
  "coordinate_size": 128,
  "detectron2_config_args": {
    "MODEL.ANCHOR_GENERATOR.SIZES": [
      [
        32
      ],
      [
        64
      ],
      [
        128
      ],
      [
        256
      ],
      [
        512
      ]
    ],
    "MODEL.BACKBONE.NAME": "build_resnet_fpn_backbone",
    "MODEL.FPN.IN_FEATURES": [
      "res2",
      "res3",
      "res4",
      "res5"
    ],
    "MODEL.MASK_ON": true,
    "MODEL.PIXEL_STD": [
      57.375,
      57.12,
      58.395
    ],
    "MODEL.POST_NMS_TOPK_TEST": 1000,
    "MODEL.RESNETS.ASPECT_RATIOS": [
      [
        0.5,
        1.0,
        2.0
      ]
    ],
    "MODEL.RESNETS.DEPTH": 101,
    "MODEL.RESNETS.NUM_GROUPS": 32,
    "MODEL.RESNETS.OUT_FEATURES": [
      "res2",
      "res3",
      "res4",
      "res5"
    ],
    "MODEL.RESNETS.SIZES": [
      [
        32
      ],
      [
        64
      ],
      [
        128
      ],
      [
        256
      ],
      [
        512
      ]
    ],
    "MODEL.RESNETS.STRIDE_IN_1X1": false,
    "MODEL.RESNETS.WIDTH_PER_GROUP": 8,
    "MODEL.ROI_BOX_HEAD.NAME": "FastRCNNConvFCHead",
    "MODEL.ROI_BOX_HEAD.NUM_FC": 2,
    "MODEL.ROI_BOX_HEAD.POOLER_RESOLUTION": 14,
    "MODEL.ROI_HEADS.IN_FEATURES": [
      "p2",
      "p3",
      "p4",
      "p5"
    ],
    "MODEL.ROI_HEADS.NAME": "StandardROIHeads",
    "MODEL.ROI_HEADS.NUM_CLASSES": 5,
    "MODEL.ROI_MASK_HEAD.NAME": "MaskRCNNConvUpsampleHead",
    "MODEL.ROI_MASK_HEAD.NUM_CONV": 4,
    "MODEL.ROI_MASK_HEAD.POOLER_RESOLUTION": 7,
    "MODEL.RPN.IN_FEATURES": [
      "p2",
      "p3",
      "p4",
      "p5",
      "p6"
    ],
    "MODEL.RPN.POST_NMS_TOPK_TRAIN": 1000,
    "MODEL.RPN.PRE_NMS_TOPK_TEST": 1000,
    "MODEL.RPN.PRE_NMS_TOPK_TRAIN": 2000
  },
  "eos_token_id": 2,
  "fast_qkv": false,
  "gradient_checkpointing": false,
  "has_relative_attention_bias": false,
  "has_spatial_attention_bias": false,
  "has_visual_segment_embedding": true,
  "hidden_act": "gelu",
  "hidden_dropout_prob": 0.1,
  "hidden_size": 768,
  "id2label": {
    "0": "B-AMTK",
    "1": "B-BADDK",
    "2": "B-BADDV",
    "3": "B-BNAMEK",
    "4": "B-BNAMEV",
    "5": "B-BTAXK",
    "6": "B-DT",
    "7": "B-NOK",
    "8": "B-SADDK",
    "9": "B-SADDV",
    "10": "B-SERIK",
    "11": "B-SERIV",
    "12": "B-SNAMEK",
    "13": "B-SNAMEV",
    "14": "B-STAXK",
    "15": "B-TOTK",
    "16": "B-VATK",
    "17": "E-AMTK",
    "18": "E-BADDK",
    "19": "E-BADDV",
    "20": "E-BNAMEK",
    "21": "E-BNAMEV",
    "22": "E-BTAXK",
    "23": "E-DT",
    "24": "E-NOK",
    "25": "E-SADDK",
    "26": "E-SADDV",
    "27": "E-SERIK",
    "28": "E-SERIV",
    "29": "E-SNAMEK",
    "30": "E-SNAMEV",
    "31": "E-STAXK",
    "32": "E-TOTK",
    "33": "E-VATK",
    "34": "I-AMTK",
    "35": "I-BADDK",
    "36": "I-BADDV",
    "37": "I-BNAMEK",
    "38": "I-BNAMEV",
    "39": "I-BTAXK",
    "40": "I-DT",
    "41": "I-NOK",
    "42": "I-SADDK",
    "43": "I-SADDV",
    "44": "I-SERIK",
    "45": "I-SNAMEK",
    "46": "I-SNAMEV",
    "47": "I-STAXK",
    "48": "I-TOTK",
    "49": "I-VATK",
    "50": "O",
    "51": "S-AMTV",
    "52": "S-BADDV",
    "53": "S-BNAMEV",
    "54": "S-BTAXV",
    "55": "S-NOK",
    "56": "S-NOV",
    "57": "S-SADDK",
    "58": "S-SERIK",
    "59": "S-SERIV",
    "60": "S-STAXK",
    "61": "S-STAXV",
    "62": "S-TOTV",
    "63": "S-VATV"
  },
  "image_feature_pool_shape": [
    7,
    7,
    256
  ],
  "initializer_range": 0.02,
  "intermediate_size": 3072,
  "label2id": {
    "B-AMTK": 0,
    "B-BADDK": 1,
    "B-BADDV": 2,
    "B-BNAMEK": 3,
    "B-BNAMEV": 4,
    "B-BTAXK": 5,
    "B-DT": 6,
    "B-NOK": 7,
    "B-SADDK": 8,
    "B-SADDV": 9,
    "B-SERIK": 10,
    "B-SERIV": 11,
    "B-SNAMEK": 12,
    "B-SNAMEV": 13,
    "B-STAXK": 14,
    "B-TOTK": 15,
    "B-VATK": 16,
    "E-AMTK": 17,
    "E-BADDK": 18,
    "E-BADDV": 19,
    "E-BNAMEK": 20,
    "E-BNAMEV": 21,
    "E-BTAXK": 22,
    "E-DT": 23,
    "E-NOK": 24,
    "E-SADDK": 25,
    "E-SADDV": 26,
    "E-SERIK": 27,
    "E-SERIV": 28,
    "E-SNAMEK": 29,
    "E-SNAMEV": 30,
    "E-STAXK": 31,
    "E-TOTK": 32,
    "E-VATK": 33,
    "I-AMTK": 34,
    "I-BADDK": 35,
    "I-BADDV": 36,
    "I-BNAMEK": 37,
    "I-BNAMEV": 38,
    "I-BTAXK": 39,
    "I-DT": 40,
    "I-NOK": 41,
    "I-SADDK": 42,
    "I-SADDV": 43,
    "I-SERIK": 44,
    "I-SNAMEK": 45,
    "I-SNAMEV": 46,
    "I-STAXK": 47,
    "I-TOTK": 48,
    "I-VATK": 49,
    "O": 50,
    "S-AMTV": 51,
    "S-BADDV": 52,
    "S-BNAMEV": 53,
    "S-BTAXV": 54,
    "S-NOK": 55,
    "S-NOV": 56,
    "S-SADDK": 57,
    "S-SERIK": 58,
    "S-SERIV": 59,
    "S-STAXK": 60,
    "S-STAXV": 61,
    "S-TOTV": 62,
    "S-VATV": 63
  },
  "layer_norm_eps": 1e-05,
  "max_2d_position_embeddings": 1024,
  "max_position_embeddings": 514,
  "max_rel_2d_pos": 256,
  "max_rel_pos": 128,
  "model_type": "layoutlmv2",
  "num_attention_heads": 12,
  "num_hidden_layers": 12,
  "output_past": true,
  "pad_token_id": 1,
  "rel_2d_pos_bins": 64,
  "rel_pos_bins": 32,
  "shape_size": 128,
  "tokenizer_class": "LayoutXLMTokenizer",
  "torch_dtype": "float32",
  "transformers_version": "4.42.3",
  "type_vocab_size": 1,
  "vocab_size": 250002
}