hfl/chinese-xlnet-base/bert4torch_config.json

{
    "model": "xlnet",
    "attn_type": "bi",
    "bi_data": false,
    "bos_token_id": 1,
    "clamp_len": -1,
    "intermediate_size": 3072,
    "hidden_size": 768,
    "hidden_dropout_prob": 0.1,
    "end_n_top": 5,
    "eos_token_id": 2,
    "hidden_act": "relu",
    "initializer_range": 0.02,
    "layer_norm_eps": 1e-12,
    "mem_len": null,
    "num_attention_heads": 12,
    "num_hidden_layers": 12,
    "output_past": true,
    "pad_token_id": 5,
    "reuse_len": null,
    "same_length": false,
    "start_n_top": 5,
    "summary_activation": "tanh",
    "summary_last_hidden_dropout_prob": 0.1,
    "summary_type": "last",
    "summary_use_proj": true,
    "untie_r": true,
    "vocab_size": 32000
  }