config.json

{
    "architectures": [
      "BertForPreTraining"
    ],
    "auto_map": {
      "AutoModel": "modeling_bert.BertForPreTraining"
    },
    "attention_probs_dropout_prob": 0.1,
    "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": 4096,
    "model_type": "bert",
    "num_attention_heads": 12,
    "num_hidden_layers": 12,
    "pad_token_id": 3,
    "pre_layer_norm": true,
    "last_layer_norm": false,
    "position_embedding_type": "rotary",
    "rotary_base": 10000,
    "rotary_dim": 32,
    "transformers_version": "4.6.0.dev0",
    "type_vocab_size": 2,
    "use_cache": true,
    "vocab_size": 32000,
    "sparse_config_cls": "deepspeed.ops.sparse_attention:BigBirdSparsityConfig",
    "sparse_attention": {
      "num_heads": 12,
      "block": 64,
      "different_layout_per_head": true,
      "num_sliding_window_blocks": 3,
      "num_global_blocks": 2,
      "num_random_blocks": 3
    }
}