config.json

{
    "_name_or_path": "togethercomputer/m2-bert-80M-8k",
    "alibi_starting_size": 8192,
    "architectures": [
        "BertForMaskedLM"
    ],
    "attention_probs_dropout_prob": 0.0,
    "bidirectional": true,
    "auto_map": {
        "AutoConfig": "configuration_bert.BertConfig",
        "AutoModelForSequenceClassification": "bert_layers.BertForMaskedLM",
        "AutoTokenizer": "bert-base-uncased"
    },
    "classifier_dropout": null,
    "gradient_checkpointing": false,
    "hidden_act": "gelu",
    "hidden_dropout_prob": 0.1,
    "hidden_size": 768,
    "hyena_emb_dim": 5,
    "hyena_filter_dropout": 0.2,
    "hyena_filter_order": 128,
    "hyena_lr_pos_emb": 1e-05,
    "hyena_training_additions": false,
    "hyena_w": 10,
    "hyena_w_mod": 1,
    "hyena_wd": 0.1,
    "initializer_range": 0.02,
    "intermediate_size": 3072,
    "layer_norm_eps": 1e-12,
    "long_conv_kernel_learning_rate": 0.001,
    "long_conv_l_max": 8192,
    "max_position_embeddings": 8192,
    "model_type": "m2_bert",
    "monarch_mlp_nblocks": 4,
    "num_attention_heads": 12,
    "num_hidden_layers": 12,
    "pad_token_id": 0,
    "pool_all": false,
    "position_embedding_type": "absolute",
    "residual_long_conv": true,
    "transformers_version": "4.28.1",
    "type_vocab_size": 2,
    "use_cache": true,
    "use_glu_mlp": true, 
    "use_monarch_mlp": true,
    "use_positional_encodings": true,
    "vocab_size": 30528
}