config.json

{
    "adaptive": true,
    "architectures": [
      "LSGRobertaForMaskedLM"
    ],
    "attention_probs_dropout_prob": 0.1,
    "auto_map": {
      "AutoConfig": "modeling_lsg_roberta.LSGRobertaConfig",
      "AutoModel": "modeling_lsg_roberta.LSGRobertaModel",
      "AutoModelForCausalLM": "modeling_lsg_roberta.LSGRobertaForCausalLM",
      "AutoModelForMaskedLM": "modeling_lsg_roberta.LSGRobertaForMaskedLM",
      "AutoModelForMultipleChoice": "modeling_lsg_roberta.LSGRobertaForMultipleChoice",
      "AutoModelForQuestionAnswering": "modeling_lsg_roberta.LSGRobertaForQuestionAnswering",
      "AutoModelForSequenceClassification": "modeling_lsg_roberta.LSGRobertaForSequenceClassification",
      "AutoModelForTokenClassification": "modeling_lsg_roberta.LSGRobertaForTokenClassification"
    },
    "base_model_prefix": "lsg",
    "block_size": 128,
    "bos_token_id": 0,
    "classifier_dropout": null,
    "eos_token_id": 2,
    "hidden_act": "gelu",
    "hidden_dropout_prob": 0.1,
    "hidden_size": 768,
    "initializer_range": 0.02,
    "intermediate_size": 3072,
    "layer_norm_eps": 1e-05,
    "lsh_num_pre_rounds": 1,
    "mask_first_token": false,
    "max_position_embeddings": 4098,
    "model_type": "roberta",
    "num_attention_heads": 12,
    "num_global_tokens": 1,
    "num_hidden_layers": 6,
    "pad_token_id": 1,
    "pool_with_global": true,
    "position_embedding_type": "absolute",
    "sparse_block_size": 128,
    "sparsity_factor": 2,
    "sparsity_type": "norm",
    "torch_dtype": "float32",
    "transformers_version": "4.30.2",
    "type_vocab_size": 1,
    "use_cache": true,
    "vocab_size": 50265
  }