"""MathBananaMind configuration for Hugging Face Transformers."""

from transformers import PretrainedConfig


class MathBananaMindConfig(PretrainedConfig):
    model_type = "mathbananamind"

    def __init__(
        self,
        vocab_size=8192,
        embedding_size=32,
        hidden_size=160,
        num_hidden_layers=9,
        num_attention_heads=5,
        num_key_value_heads=1,
        intermediate_size=480,
        max_position_embeddings=1024,
        rope_theta=10000.0,
        rms_norm_eps=1e-5,
        attention_dropout=0.0,
        bos_token_id=0,
        eos_token_id=0,
        pad_token_id=1,
        unk_token_id=2,
        tie_word_embeddings=False,
        use_cache=False,
        **kwargs,
    ):
        self.vocab_size = vocab_size
        self.embedding_size = embedding_size
        self.hidden_size = hidden_size
        self.num_hidden_layers = num_hidden_layers
        self.num_attention_heads = num_attention_heads
        self.num_key_value_heads = num_key_value_heads
        self.intermediate_size = intermediate_size
        self.max_position_embeddings = max_position_embeddings
        self.rope_theta = rope_theta
        self.rms_norm_eps = rms_norm_eps
        self.attention_dropout = attention_dropout
        super().__init__(
            bos_token_id=bos_token_id,
            eos_token_id=eos_token_id,
            pad_token_id=pad_token_id,
            unk_token_id=unk_token_id,
            tie_word_embeddings=tie_word_embeddings,
            use_cache=use_cache,
            **kwargs,
        )

    @property
    def head_dim(self):
        return self.hidden_size // self.num_attention_heads