config.json

{
  "architectures": [
    "NewBertForSequenceClassification"
  ],
  "attention_probs_dropout_prob": 0.1,
  "decompose_qk": false,
  "decompose_vo": false,
  "do_distill": true,
  "do_emb_distill": false,
  "do_layer_distill": true,
  "do_mha_distill": false,
  "do_mha_layer_distill": false,
  "finetuning_task": "sst-2",
  "gradient_checkpointing": false,
  "hidden_act": "gelu",
  "hidden_dropout_prob": 0.1,
  "hidden_size": 768,
  "initializer_range": 0.02,
  "intermediate_size": 3072,
  "layer_norm_eps": 1e-12,
  "max_position_embeddings": 512,
  "model_type": "bert",
  "num_attention_heads": 12,
  "num_hidden_layers": 12,
  "output_attentions": true,
  "output_hidden_states": true,
  "pad_token_id": 0,
  "pruned_heads": {
    "0": [
      7
    ],
    "1": [
      2,
      3,
      6
    ],
    "2": [
      2,
      3,
      4,
      7,
      10
    ],
    "3": [
      3,
      4,
      7
    ],
    "4": [
      1,
      3,
      4,
      6,
      8,
      9
    ],
    "5": [
      0,
      1,
      2,
      3,
      5,
      6,
      7,
      8,
      10,
      11
    ],
    "6": [
      0,
      1,
      2,
      3,
      9
    ],
    "7": [
      5,
      6,
      7,
      9,
      11
    ],
    "8": [
      0,
      11,
      3,
      4
    ],
    "9": [
      1,
      2,
      10,
      7
    ],
    "10": [
      7
    ],
    "11": [
      0,
      8,
      3,
      4
    ]
  },
  "qk_denominator": "ori",
  "sephidden_pruned": false,
  "transform_embedding": false,
  "type_vocab_size": 2,
  "vocab_size": 30522
}