config.json

{
    "architectures": [
        "CustomElectraForPreTraining"
    ],
    "attention_probs_dropout_prob": 0.1,
    "attention_window": [
        256,
        256,
        256,
        256,
        256,
        256,
        256,
        256,
        256,
        256,
        256,
        256
    ],
    "bos_token_id": 0,
    "classifier_dropout": null,
    "embedding_size": 768,
    "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-12,
    "max_position_embeddings": 4098,
    "num_attention_heads": 12,
    "num_hidden_layers": 12,
    "pad_token_id": 1,
    "position_embedding_type": "absolute",
    "sep_token_id": 2,
    "torch_dtype": "float32",
    "transformers_version": "4.21.0",
    "type_vocab_size": 2,
    "use_cache": true,
    "vocab_size": 64000,
    "model_type": "longformer"
}