config.json

{
  "_name_or_path": "meta-llama/Llama-3.2-1B",
  "architectures": [
    "LlamaForTokenClassification"
  ],
  "attention_bias": false,
  "attention_dropout": 0.0,
  "bos_token_id": 128000,
  "eos_token_id": 128001,
  "head_dim": 64,
  "hidden_act": "silu",
  "hidden_size": 2048,
  "id2label": {
    "0": "O",
    "1": "Ti\u1ec1n c\u1ee5 th\u1ec3",
    "2": "M\u00e3 \u0111\u01a1n",
    "3": "S\u1ed1 c\u00f4ng",
    "4": "L\u01b0\u01a1ng",
    "5": "V\u1ecb tr\u00ed",
    "6": "S\u1ed1 \u0111i\u1ec3m",
    "7": "C\u00f4ng",
    "8": "Ti\u1ec1n tr\u1eebu t\u01b0\u1ee3ng",
    "9": "S\u1ed1 \u0111\u01a1n",
    "10": "Ph\u1ee5 c\u1ea5p",
    "11": "Th\u1ee9",
    "12": "S\u1ed1 gi\u1edd",
    "13": "Kho\u1ea3ng th\u1eddi gian",
    "14": "Th\u00f4ng tin CTT",
    "15": "Th\u00e1ng tr\u1eebu t\u01b0\u1ee3ng",
    "16": "Kho",
    "17": "H\u00ecnh th\u1ee9c l\u00e0m vi\u1ec7c",
    "18": "\u0110\u1ee3t",
    "19": "T\u1ef7 l\u1ec7",
    "20": "Gi\u1ea5y t\u1edd",
    "21": "M\u00e3 s\u1ed1 thu\u1ebf"
  },
  "initializer_range": 0.02,
  "intermediate_size": 8192,
  "label2id": {
    "C\u00f4ng": 7,
    "Gi\u1ea5y t\u1edd": 20,
    "H\u00ecnh th\u1ee9c l\u00e0m vi\u1ec7c": 17,
    "Kho": 16,
    "Kho\u1ea3ng th\u1eddi gian": 13,
    "L\u01b0\u01a1ng": 4,
    "M\u00e3 s\u1ed1 thu\u1ebf": 21,
    "M\u00e3 \u0111\u01a1n": 2,
    "O": 0,
    "Ph\u1ee5 c\u1ea5p": 10,
    "S\u1ed1 c\u00f4ng": 3,
    "S\u1ed1 gi\u1edd": 12,
    "S\u1ed1 \u0111i\u1ec3m": 6,
    "S\u1ed1 \u0111\u01a1n": 9,
    "Th\u00e1ng tr\u1eebu t\u01b0\u1ee3ng": 15,
    "Th\u00f4ng tin CTT": 14,
    "Th\u1ee9": 11,
    "Ti\u1ec1n c\u1ee5 th\u1ec3": 1,
    "Ti\u1ec1n tr\u1eebu t\u01b0\u1ee3ng": 8,
    "T\u1ef7 l\u1ec7": 19,
    "V\u1ecb tr\u00ed": 5,
    "\u0110\u1ee3t": 18
  },
  "max_position_embeddings": 131072,
  "mlp_bias": false,
  "model_type": "llama",
  "num_attention_heads": 32,
  "num_hidden_layers": 16,
  "num_key_value_heads": 8,
  "pretraining_tp": 1,
  "rms_norm_eps": 1e-05,
  "rope_scaling": {
    "factor": 32.0,
    "high_freq_factor": 4.0,
    "low_freq_factor": 1.0,
    "original_max_position_embeddings": 8192,
    "rope_type": "llama3"
  },
  "rope_theta": 500000.0,
  "tie_word_embeddings": true,
  "torch_dtype": "float32",
  "transformers_version": "4.44.0",
  "use_cache": true,
  "vocab_size": 128256
}