{
  "adapter_dim": 0,
  "architectures": [
    "LeanAlbertForPreTraining"
  ],
  "attention_probs_dropout_prob": 0,
  "attn_qkv_bias": true,
  "block_size": 0,
  "bos_token_id": 2,
  "classifier_dropout_prob": 0.1,
  "embedding_size": 128,
  "eos_token_id": 3,
  "hidden_act": "gelu_fused",
  "hidden_act_gated": true,
  "hidden_dropout_prob": 0,
  "hidden_size": 4096,
  "initializer_range": 0.02,
  "inner_group_num": 1,
  "intermediate_size": 16384,
  "layer_norm_eps": 1e-12,
  "lowrank_dim": 0,
  "max_position_embeddings": 512,
  "num_attention_heads": 64,
  "num_hidden_groups": 32,
  "num_hidden_layers": 32,
  "num_inner_groups": 1,
  "num_inner_matrices": 1,
  "num_memory_blocks": 0,
  "num_shared_matrices": 1,
  "out_proj_bias": true,
  "pad_token_id": 0,
  "position_embedding_type": "rotary",
  "rotary_embedding_base": 10000,
  "sandwich_norm": false,
  "share_large_matrices": true,
  "torch_dtype": "float32",
  "total_num_layer_groups": 32,
  "total_shared_matrix_sets": 1,
  "transformers_version": "4.24.0",
  "type_vocab_size": 2,
  "vocab_size": 50005
}