#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
#           This file was automatically generated from src/transformers/models/mimo_v2_flash/modular_mimo_v2_flash.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_mimo_v2_flash.py file directly. One of our CI enforces this.
#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
# Copyright 2026 Xiaomi Corporation and the HuggingFace Inc. 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 huggingface_hub.dataclasses import strict

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


@auto_docstring(checkpoint="XiaomiMiMo/MiMo-V2-Flash")
@strict
class MiMoV2FlashConfig(PreTrainedConfig):
    r"""
    n_group (`int`, *optional*, defaults to 1):
        Number of expert groups for group-based top-k routing.
    topk_group (`int`, *optional*, defaults to 1):
        Number of groups selected per token in group-based top-k routing.
    head_dim (`int`, *optional*, defaults to 192):
        Dimension of query and key heads.
    v_head_dim (`int`, *optional*, defaults to 128):
        Dimension of value heads (special case because MiMo uses a smaller v head dim than (qk) head dim )
    mlp_layer_types (`list`, *optional*):
        MLP pattern for each layer (`"dense"` or `"sparse"`). Defaults to 1 dense + rest sparse.
    attention_value_scale (`float`, *optional*, defaults to 0.707 (which is the decimal approximation
        of `sqrt(hidden_size / (num_attention_heads * v_head_dim))`):
        Constant multiplier applied to rescale the attention values.
    """

    model_type = "mimo_v2_flash"
    keys_to_ignore_at_inference = ["past_key_values"]

    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.*.self_attn.sinks": "colwise",
        "layers.*.mlp.gate_proj": "colwise",
        "layers.*.mlp.up_proj": "colwise",
        "layers.*.mlp.down_proj": "rowwise",
        "layers.*.mlp.experts.gate_up_proj": "packed_colwise",
        "layers.*.mlp.experts.down_proj": "rowwise",
        "layers.*.mlp.experts": "moe_tp_experts",
    }
    base_model_pp_plan = {
        "embed_tokens": (["input_ids"], ["inputs_embeds"]),
        "layers": (["hidden_states", "attention_mask"], ["hidden_states"]),
        "norm": (["hidden_states"], ["hidden_states"]),
    }
    base_model_ep_plan = {
        "layers.*.mlp.gate": "ep_router",
        "layers.*.mlp.experts.gate_up_proj": "grouped_gemm",
        "layers.*.mlp.experts.down_proj": "grouped_gemm",
        "layers.*.mlp.experts": "moe_tp_experts",
    }
    attribute_map = {"num_local_experts": "n_routed_experts"}

    # Overrides from Glm4MoeConfig
    vocab_size: int = 152576
    hidden_size: int = 4096
    intermediate_size: int = 16384
    num_hidden_layers: int = 48
    num_attention_heads: int = 64
    num_key_value_heads: int = 4
    hidden_act: str = "silu"
    max_position_embeddings: int = 131072
    initializer_range: float = 0.02
    rms_norm_eps: float = 1e-5
    use_cache: bool = True
    tie_word_embeddings: bool = False
    rope_parameters: RopeParameters | dict | None = None
    attention_bias: bool = False
    attention_dropout: float | int = 0.0
    moe_intermediate_size: int = 2048
    num_experts_per_tok: int = 8
    n_routed_experts: int = 256
    routed_scaling_factor: float | None = 1.0
    n_group: int = 1
    topk_group: int = 1
    norm_topk_prob: bool = True
    bos_token_id: int | None = 1
    eos_token_id: int | list[int] | None = None
    pad_token_id: int | None = None
    # MiMo-V2-Flash specific
    head_dim: int = 192
    v_head_dim: int = 128
    sliding_window: int = 128
    layer_types: list[str] | None = None
    mlp_layer_types: list[str] | None = None
    attention_value_scale: float | None = 0.707

    def __post_init__(self, **kwargs):
        # Full attention for the first layer and every 6th layer; SWA for the rest.
        if self.layer_types is None:
            self.layer_types = [
                "full_attention" if (i == 0 or not ((i + 1) % 6)) else "sliding_attention"
                for i in range(self.num_hidden_layers)
            ]
        # First layer is a dense MLP, the rest are MoE.
        if self.mlp_layer_types is None:
            self.mlp_layer_types = ["dense"] + ["sparse"] * (self.num_hidden_layers - 1)
        # Per-layer rope defaults matching the XiaomiMiMo/MiMo-V2-Flash pretrained thetas.
        if self.rope_parameters is None:
            self.rope_parameters = {
                "full_attention": {"rope_type": "default", "rope_theta": 5_000_000.0, "partial_rotary_factor": 0.334},
                "sliding_attention": {"rope_type": "default", "rope_theta": 10_000.0, "partial_rotary_factor": 0.334},
            }
        # BC: The hub config.json stores `routed_scaling_factor` as null
        if self.routed_scaling_factor is None:
            self.routed_scaling_factor = 1.0

        super().__post_init__(**kwargs)

    def convert_rope_params_to_dict(self, **kwargs):
        return kwargs


__all__ = ["MiMoV2FlashConfig"]