config.json

{
  "_name_or_path": "vinai/phobert-large",
  "architectures": [
    "RobertaForTokenClassification"
  ],
  "attention_probs_dropout_prob": 0.1,
  "bos_token_id": 0,
  "classifier_dropout": null,
  "eos_token_id": 2,
  "gradient_checkpointing": false,
  "hidden_act": "gelu",
  "hidden_dropout_prob": 0.1,
  "hidden_size": 1024,
  "id2label": {
    "0": "M\u00e3 \u0111\u01a1n",
    "1": "Kh\u1ed1i l\u01b0\u1ee3ng",
    "2": "STK",
    "3": "\u0110\u01a1n v\u1ecb \u0111o",
    "4": "O",
    "5": "S\u0110T",
    "6": "S\u1ea3n ph\u1ea9m c\u1ee5 th\u1ec3",
    "7": "Ti\u1ec1n",
    "8": "CCCD - CMT",
    "9": "T\u00ean ng\u01b0\u1eddi",
    "10": "S\u1ea3n ph\u1ea9m tr\u1eebu t\u01b0\u1ee3ng",
    "11": "Ph\u01b0\u01a1ng th\u1ee9c thanh to\u00e1n",
    "12": "Ca",
    "13": "Ng\u00e2n h\u00e0ng",
    "14": "Ng\u00e0y",
    "15": "\u0110\u1ecba ch\u1ec9 c\u1ee5 th\u1ec3",
    "16": "Gi\u1edd",
    "17": "Email",
    "18": "Ng\u00e0y tr\u1eebu t\u01b0\u1ee3ng",
    "19": "\u0110\u1ecba ch\u1ec9 tr\u1eebu t\u01b0\u1ee3ng",
    "20": "S\u0110T tr\u1eebu t\u01b0\u1ee3ng"
  },
  "initializer_range": 0.02,
  "intermediate_size": 4096,
  "label2id": {
    "CCCD - CMT": 8,
    "Ca": 12,
    "Email": 17,
    "Gi\u1edd": 16,
    "Kh\u1ed1i l\u01b0\u1ee3ng": 1,
    "M\u00e3 \u0111\u01a1n": 0,
    "Ng\u00e0y": 14,
    "Ng\u00e0y tr\u1eebu t\u01b0\u1ee3ng": 18,
    "Ng\u00e2n h\u00e0ng": 13,
    "O": 4,
    "Ph\u01b0\u01a1ng th\u1ee9c thanh to\u00e1n": 11,
    "STK": 2,
    "S\u0110T": 5,
    "S\u0110T tr\u1eebu t\u01b0\u1ee3ng": 20,
    "S\u1ea3n ph\u1ea9m c\u1ee5 th\u1ec3": 6,
    "S\u1ea3n ph\u1ea9m tr\u1eebu t\u01b0\u1ee3ng": 10,
    "Ti\u1ec1n": 7,
    "T\u00ean ng\u01b0\u1eddi": 9,
    "\u0110\u01a1n v\u1ecb \u0111o": 3,
    "\u0110\u1ecba ch\u1ec9 c\u1ee5 th\u1ec3": 15,
    "\u0110\u1ecba ch\u1ec9 tr\u1eebu t\u01b0\u1ee3ng": 19
  },
  "layer_norm_eps": 1e-05,
  "max_position_embeddings": 258,
  "model_type": "roberta",
  "num_attention_heads": 16,
  "num_hidden_layers": 24,
  "pad_token_id": 1,
  "position_embedding_type": "absolute",
  "tokenizer_class": "PhobertTokenizer",
  "torch_dtype": "float32",
  "transformers_version": "4.43.1",
  "type_vocab_size": 1,
  "use_cache": true,
  "vocab_size": 64001
}