#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
#           This file was automatically generated from src/transformers/models/hy_v3/modular_hy_v3.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_hy_v3.py file directly. One of our CI enforces this.
#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
# Copyright 2026 Tencent HunYuan Team 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="tencent/Hy3-preview")
@strict
class HYV3Config(PreTrainedConfig):
    r"""
    router_scaling_factor (*float*):
        Scaling factor on the top-k weighs of the MoE expert selection.
    enable_moe_fp32_combine (*bool*):
        Whether to add the shared experts to the final MoE result in fp32 or the base existing dtype of the model.
    mlp_layer_types (`list`, *optional*):
        MLP (Moe vs Dense) pattern for each layer.

    Example:
        ```python
        >>> from transformers import HYV3Config, HYV3Model

        >>> config = HYV3Config()
        >>> model = HYV3Model(config)
        ```
    """

    model_type = "hy_v3"
    default_theta = 11_158_840.0
    keys_to_ignore_at_inference = ["past_key_values"]
    attribute_map = {
        "num_local_experts": "num_experts",
    }
    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.q_norm": "replicated_with_grad_allreduce",
        "layers.*.self_attn.k_norm": "replicated_with_grad_allreduce",
        "layers.*.self_attn.o_proj": "rowwise",
        "layers.*.mlp.experts.gate_up_proj": "packed_colwise",
        "layers.*.mlp.experts.down_proj": "rowwise",
        "layers.*.mlp.experts": "moe_tp_experts",
        "layers.*.mlp.shared_experts.gate_proj": "colwise",
        "layers.*.mlp.shared_experts.up_proj": "colwise",
        "layers.*.mlp.shared_experts.down_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 = 120832
    hidden_size: int = 4096
    intermediate_size: int = 13312
    num_hidden_layers: int = 80
    num_attention_heads: int = 64
    num_key_value_heads: int = 8
    head_dim: int = 128
    hidden_act: str = "silu"
    max_position_embeddings: int = 131072
    initializer_range: float = 0.006
    rms_norm_eps: float = 1e-5
    use_cache: bool = True
    pad_token_id: int | None = None
    bos_token_id: int | None = None
    eos_token_id: int | list[int] | None = None
    tie_word_embeddings: bool = False
    attention_bias: bool = False
    attention_dropout: float = 0.0
    mlp_bias: bool = False
    num_experts: int | None = 192
    num_experts_per_tok: int | None = 8
    num_shared_experts: int | None = 1
    moe_intermediate_size: int = 1536
    router_scaling_factor: float = 2.826
    enable_moe_fp32_combine: bool = True
    mlp_layer_types: list[str] | None = None
    output_router_logits: bool = False
    rope_parameters: RopeParameters | dict | None = None

    def __post_init__(self, **kwargs):
        if self.mlp_layer_types is None:
            self.mlp_layer_types = ["dense"] * (1 if self.num_hidden_layers > 0 else 0) + ["sparse"] * max(
                self.num_hidden_layers - 1, 0
            )

        super().__post_init__(**kwargs)


__all__ = ["HYV3Config"]