config.json

{
  "_name_or_path": "/mnt/msranlpintern/models/markuplm-large_500k_ffn_scale-as-roberta_lr5e-5/checkpoint-240000/",
  "architectures": [
    "MarkupLMForQuestionAnswering"
  ],
  "attention_probs_dropout_prob": 0.1,
  "bos_token_id": 0,
  "classifier_dropout": null,
  "disable_xpath_emb": false,
  "enable_xpath_emb_normalize": false,
  "eos_token_id": 2,
  "ffn_proj": true,
  "fix_xpath_proj": false,
  "gradient_checkpointing": false,
  "has_relative_attention_bias": false,
  "has_tree_attention_bias": false,
  "hidden_act": "gelu",
  "hidden_dropout_prob": 0.1,
  "hidden_size": 1024,
  "init_model_type": "roberta",
  "initializer_range": 0.02,
  "intermediate_size": 4096,
  "layer_norm_eps": 1e-05,
  "max_depth": 50,
  "max_position_embeddings": 514,
  "max_rel_pos": 128,
  "max_tree_id_unit_embeddings": 1024,
  "max_xpath_subs_unit_embeddings": 1024,
  "max_xpath_tag_unit_embeddings": 256,
  "model_type": "markuplm",
  "num_attention_heads": 16,
  "num_hidden_layers": 24,
  "pad_token_id": 1,
  "pos_mode_for_path_emb": 1,
  "position_embedding_type": "absolute",
  "rel_pos_bins": 32,
  "torch_dtype": "float32",
  "transformers_version": "4.10.2",
  "tree_id_unit_hidden_size": 32,
  "tree_max_rel_pos": 128,
  "tree_rel_pos_bins": 32,
  "type_vocab_size": 1,
  "use_cache": true,
  "vocab_size": 50267,
  "xpath_unit_hidden_size": 32
}