#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
#           This file was automatically generated from src/transformers/models/ultrabert/modular_ultrabert.py.
#               Do NOT edit this file manually as any edits will be overwritten by the generation of
#             the file from the modular. If any change should be done, please apply the change to the
#                          modular_ultrabert.py file directly. One of our CI enforces this.
#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
# Copyright 2026 the HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License").
# You may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

from typing import Literal

from ...configuration_utils import PreTrainedConfig, strict
from ...modeling_rope_utils import RopeParameters
from ...utils import auto_docstring


@auto_docstring(checkpoint="dmis-lab/UltraBERT-150M")
@strict
class UltrabertConfig(PreTrainedConfig):
    r"""
    sliding_window (`int`, *optional*, defaults to 128):
        Inclusive attention radius on either side of each token in sliding attention layers.
    prediction_head_size (`int`, *optional*, defaults to 768):
        Intermediate dimension of the two-layer masked language modeling prediction head.
    classifier_pooling (`str`, *optional*, defaults to `"cls"`):
        Pooling method for sequence classification. Either `"cls"` or `"mean"`.
    """

    model_type = "ultrabert"
    keys_to_ignore_at_inference = ["past_key_values"]
    # Default tensor parallel plan for base model `UltrabertModel`
    base_model_tp_plan = {
        "layers.*.self_attn.q_proj": "colwise",
        "layers.*.self_attn.k_proj": "colwise",
        "layers.*.self_attn.v_proj": "colwise",
        "layers.*.self_attn.o_proj": "rowwise",
        "layers.*.mlp.gate_proj": "colwise",
        "layers.*.mlp.up_proj": "colwise",
        "layers.*.mlp.down_proj": "rowwise",
    }
    base_model_pp_plan = {
        "embed_tokens": (["input_ids"], ["inputs_embeds"]),
        "layers": (["hidden_states", "attention_mask"], ["hidden_states"]),
        "norm": (["hidden_states"], ["hidden_states"]),
    }

    vocab_size: int = 32000
    hidden_size: int = 4096
    intermediate_size: int = 11008
    num_hidden_layers: int = 32
    num_attention_heads: int = 32
    num_key_value_heads: int | None = None
    hidden_act: str = "silu"
    max_position_embeddings: int = 32768
    initializer_range: float = 0.02
    rms_norm_eps: float = 1e-6
    use_cache: bool = False
    pad_token_id: int | None = 0
    bos_token_id: int | None = 0
    eos_token_id: int | list[int] | None = 0
    tie_word_embeddings: bool = False
    rope_parameters: RopeParameters | dict | None = None
    attention_bias: bool = False
    attention_dropout: int | float = 0.0
    mlp_bias: bool = False
    head_dim: int | None = None
    classifier_pooling: Literal["cls", "mean"] = "cls"
    sliding_window: int = 128
    layer_types: list[str] | None = None
    prediction_head_size: int = 768
    classifier_dropout: float = 0.0

    def __post_init__(self, **kwargs):
        # BC -> existing checkpoints store attention windows per layer.
        if "attention_windows" in kwargs:
            attention_windows = kwargs.pop("attention_windows")
            self.sliding_window = next((window for window in attention_windows if window), self.sliding_window)
            if self.layer_types is None:
                self.layer_types = [
                    "sliding_attention" if window else "full_attention" for window in attention_windows
                ]
        if self.layer_types is None:
            self.layer_types = [
                "sliding_attention" if (i + 1) % 3 else "full_attention" for i in range(self.num_hidden_layers)
            ]

        # BC -> existing checkpoints store RoPE bases in rope_theta.
        rope_theta = kwargs.pop("rope_theta", None)
        if rope_theta is not None and self.rope_parameters is None:
            self.rope_parameters = {
                kind: {"rope_type": "default", "rope_theta": theta}
                for kind, theta in zip(self.layer_types, rope_theta)
            }
        if self.rope_parameters is None:
            self.rope_parameters = {
                "sliding_attention": {"rope_type": "default", "rope_theta": 10_000.0},
                "full_attention": {"rope_type": "default", "rope_theta": 2_560_000.0},
            }
        if self.num_key_value_heads is None:
            self.num_key_value_heads = self.num_attention_heads
        if self.head_dim is None:
            self.head_dim = self.hidden_size // self.num_attention_heads
        if self.num_key_value_heads is None:
            self.num_key_value_heads = self.num_attention_heads

        super().__post_init__(**kwargs)

    def validate_architecture(self):
        """Part of `@strict`-powered validation. Validates the architecture of the config."""
        if self.hidden_size % self.num_attention_heads != 0:
            raise ValueError(
                f"The hidden size ({self.hidden_size}) is not a multiple of the number of attention "
                f"heads ({self.num_attention_heads})."
            )


__all__ = ["UltrabertConfig"]
