configuration_lightning.py

from transformers import PretrainedConfig


class LightningConfig(PretrainedConfig):

    model_type = "lightning"

    def __init__(
        self,
        vocab_size=50000,
        d_model=256,
        nhead=4,
        num_layers=4,
        dropout=0.1,
        max_seq_len=160,
        pad_token_id=0,
        eos_token_id=2,
        **kwargs
    ):
        super().__init__(
            tie_word_embeddings=False,
            pad_token_id=pad_token_id,
            eos_token_id=eos_token_id,
            **kwargs
        )

        self.vocab_size = vocab_size
        self.d_model = d_model
        self.nhead = nhead
        self.num_layers = num_layers
        self.dropout = dropout
        self.num_hidden_layers = num_layers
        self.num_attention_heads = nhead
        self.hidden_size = d_model
        self.max_seq_len = max_seq_len