from transformers import PretrainedConfig


class CALIConfig(PretrainedConfig):
    model_type = "cali"

    def __init__(
        self,
        vocab_size: int = 32000,
        hidden_dim: int = 768,
        num_layers: int = 11,
        num_heads: int = 4,
        num_kv_heads: int = 1,
        head_dim: int = 192,
        seq_len: int = 1024,
        ffn_multiplier: float = 3.0,
        dropout: float = 0.0,
        tie_embeddings: bool = True,
        rope_theta: float = 10000.0,
        rms_norm_eps: float = 1e-6,
        initializer_range: float = 0.02,
        use_cache: bool = True,
        pad_token_id: int = None,
        bos_token_id: int = 1,
        eos_token_id: int = 2,
        **kwargs,
    ):
        self.vocab_size = vocab_size
        self.hidden_dim = hidden_dim
        self.num_layers = num_layers
        self.num_heads = num_heads
        self.num_kv_heads = num_kv_heads
        self.head_dim = head_dim
        self.seq_len = seq_len
        self.ffn_multiplier = ffn_multiplier
        self.dropout = dropout
        self.tie_embeddings = tie_embeddings
        self.rope_theta = rope_theta
        self.rms_norm_eps = rms_norm_eps
        self.initializer_range = initializer_range
        self.use_cache = use_cache

        super().__init__(
            pad_token_id=pad_token_id,
            bos_token_id=bos_token_id,
            eos_token_id=eos_token_id,
            **kwargs,
        )