config.json

{
  "_name_or_path": "dicta-il/dictabert",
  "architectures": [
    "BertForMultiLabelTokenClassification"
  ],
  "attention_probs_dropout_prob": 0.1,
  "classifier_dropout": null,
  "gradient_checkpointing": false,
  "hidden_act": "gelu",
  "hidden_dropout_prob": 0.1,
  "hidden_size": 768,
  "id2label": {
    "0": "\"_",
    "1": "_\u05d4",
    "2": "_\u05d4\u05d5",
    "3": "_\u05d4\u05dd",
    "4": "_\u05d4\u05df",
    "5": "_\u05d5",
    "6": "_\u05d9",
    "7": "_\u05d9\u05d4",
    "8": "_\u05d9\u05d5",
    "9": "_\u05da",
    "10": "_\u05db\u05dd",
    "11": "_\u05dd",
    "12": "_\u05df",
    "13": "_\u05e0\u05d5",
    "14": "_\u05e0\u05d9",
    "15": "'_",
    "16": "\u05d1_",
    "17": "\u05d4_",
    "18": "\u05d5_",
    "19": "\u05db_",
    "20": "\u05db\u05e9_",
    "21": "\u05dc_",
    "22": "\u05de_",
    "23": "\u05de\u05e9_",
    "24": "\u05e9_"
  },
  "initializer_range": 0.02,
  "intermediate_size": 3072,
  "label2id": {
    "\"_": 0,
    "'_": 15,
    "_\u05d4": 1,
    "_\u05d4\u05d5": 2,
    "_\u05d4\u05dd": 3,
    "_\u05d4\u05df": 4,
    "_\u05d5": 5,
    "_\u05d9": 6,
    "_\u05d9\u05d4": 7,
    "_\u05d9\u05d5": 8,
    "_\u05da": 9,
    "_\u05db\u05dd": 10,
    "_\u05dd": 11,
    "_\u05df": 12,
    "_\u05e0\u05d5": 13,
    "_\u05e0\u05d9": 14,
    "\u05d1_": 16,
    "\u05d4_": 17,
    "\u05d5_": 18,
    "\u05db_": 19,
    "\u05db\u05e9_": 20,
    "\u05dc_": 21,
    "\u05de_": 22,
    "\u05de\u05e9_": 23,
    "\u05e9_": 24
  },
  "layer_norm_eps": 1e-12,
  "max_position_embeddings": 512,
  "model_type": "bert",
  "newmodern": true,
  "num_attention_heads": 12,
  "num_hidden_layers": 12,
  "pad_token_id": 3,
  "position_embedding_type": "absolute",
  "torch_dtype": "float32",
  "transformers_version": "4.44.2",
  "type_vocab_size": 2,
  "use_cache": true,
  "vocab_size": 128000
}