modeling_mimo_v2_flash.py

# coding=utf-8
#
# Copyright 2025 Xiaomi Corporation.
# Copyright 2025 The HuggingFace Inc. team.
#
# 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 Callable, Optional, Tuple, Union

import torch
import torch.nn as nn
import torch.nn.functional as F

from transformers.generation import GenerationMixin
from transformers.activations import ACT2FN
from transformers.cache_utils import Cache, DynamicCache
from transformers.integrations import use_kernel_forward_from_hub

from transformers.modeling_outputs import (
    BaseModelOutputWithPast,
    CausalLMOutputWithPast,
)

from transformers.masking_utils import create_causal_mask, create_sliding_window_causal_mask
from transformers.modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update
from transformers.modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel
from transformers.processing_utils import Unpack
from transformers.utils import (
    logging,
)

from transformers.modeling_outputs import MoeModelOutputWithPast
from transformers.utils import TransformersKwargs, auto_docstring, can_return_tuple
from .configuration_mimo_v2_flash import MiMoV2FlashConfig

logger = logging.get_logger(__name__)


def rotate_half(x):
    """Rotates half the hidden dims of the input."""
    x1 = x[..., : x.shape[-1] // 2]
    x2 = x[..., x.shape[-1] // 2:]
    return torch.cat((-x2, x1), dim=-1)


def apply_rotary_pos_emb(q, k, cos, sin, position_ids=None, unsqueeze_dim=1):
    """Applies Rotary Position Embedding to the query and key tensors.

    Args:
        q (`torch.Tensor`): The query tensor.
        k (`torch.Tensor`): The key tensor.
        cos (`torch.Tensor`): The cosine part of the rotary embedding.
        sin (`torch.Tensor`): The sine part of the rotary embedding.
        position_ids (`torch.Tensor`, *optional*):
            Deprecated and unused.
        unsqueeze_dim (`int`, *optional*, defaults to 1):
            The 'unsqueeze_dim' argument specifies the dimension along which to unsqueeze cos[position_ids] and
            sin[position_ids] so that they can be properly broadcasted to the dimensions of q and k. For example, note
            that cos[position_ids] and sin[position_ids] have the shape [batch_size, seq_len, head_dim]. Then, if q and
            k have the shape [batch_size, heads, seq_len, head_dim], then setting unsqueeze_dim=1 makes
            cos[position_ids] and sin[position_ids] broadcastable to the shapes of q and k. Similarly, if q and k have
            the shape [batch_size, seq_len, heads, head_dim], then set unsqueeze_dim=2.
    Returns:
        `tuple(torch.Tensor)` comprising of the query and key tensors rotated using the Rotary Position Embedding.
    """
    cos = cos.unsqueeze(unsqueeze_dim)
    sin = sin.unsqueeze(unsqueeze_dim)
    q_embed = (q * cos) + (rotate_half(q) * sin)
    k_embed = (k * cos) + (rotate_half(k) * sin)
    return q_embed, k_embed


def repeat_kv(hidden_states: torch.Tensor, n_rep: int) -> torch.Tensor:
    """
    This is the equivalent of torch.repeat_interleave(x, dim=1, repeats=n_rep). The hidden states go from (batch,
    num_key_value_heads, seqlen, head_dim) to (batch, num_attention_heads, seqlen, head_dim)
    """
    batch, num_key_value_heads, slen, head_dim = hidden_states.shape
    if n_rep == 1:
        return hidden_states
    hidden_states = hidden_states[:, :, None, :, :].expand(batch, num_key_value_heads, n_rep, slen, head_dim)
    return hidden_states.reshape(batch, num_key_value_heads * n_rep, slen, head_dim)


def eager_attention_forward(
    module: nn.Module,
    query: torch.Tensor,
    key: torch.Tensor,
    value: torch.Tensor,
    attention_mask: Optional[torch.Tensor],
    scaling: float,
    dropout: float = 0.0,
    sinks: Optional[torch.Tensor] = None,
):
    key_states = repeat_kv(key, module.num_key_value_groups)
    value_states = repeat_kv(value, module.num_key_value_groups)
    attn_weights = torch.matmul(query, key_states.transpose(2, 3)) * scaling
    if attention_mask is not None:
        causal_mask = attention_mask[:, :, :, : key_states.shape[-2]]
        attn_weights = attn_weights + causal_mask

    if sinks is not None:
        sinks = module.attention_sink_bias.reshape(1, -1, 1, 1).expand(query.shape[0], -1, query.shape[-2], -1)
        attn_weights = torch.cat([attn_weights, sinks], dim=-1)

    attn_weights = attn_weights - attn_weights.max(dim=-1, keepdim=True).values
    probs = F.softmax(attn_weights, dim=-1, dtype=attn_weights.dtype)

    if sinks is not None:
        probs = probs[..., :-1]  # we drop the sink here

    attn_weights = nn.functional.dropout(probs, p=dropout, training=module.training)
    attn_output = torch.matmul(attn_weights, value_states)
    attn_output = attn_output.transpose(1, 2).contiguous()
    return attn_output, attn_weights


@use_kernel_forward_from_hub("RMSNorm")
class MiMoV2RMSNorm(nn.Module):
    def __init__(self, hidden_size, eps=1e-6):
        """
            MiMoV2RMSNorm is equivalent to T5LayerNorm
        """
        super().__init__()
        self.weight = nn.Parameter(torch.ones(hidden_size))
        self.variance_epsilon = eps

    def forward(self, hidden_states):
        input_dtype = hidden_states.dtype
        hidden_states = hidden_states.to(torch.float32)
        variance = hidden_states.pow(2).mean(-1, keepdim=True)
        hidden_states = hidden_states * torch.rsqrt(variance + self.variance_epsilon)
        return self.weight * hidden_states.to(input_dtype)


class MiMoV2MLP(nn.Module):
    """MiMoV2MLP matching the gate, up, and down projection layers."""

    def __init__(self, config: MiMoV2FlashConfig, intermediate_size=None):
        super().__init__()
        self.config = config
        self.hidden_size = config.hidden_size
        self.intermediate_size = config.intermediate_size if intermediate_size is None else intermediate_size
        self.gate_proj = nn.Linear(self.hidden_size, self.intermediate_size, bias=False)
        self.up_proj = nn.Linear(self.hidden_size, self.intermediate_size, bias=False)
        self.down_proj = nn.Linear(self.intermediate_size, self.hidden_size, bias=False)
        self.act_fn = ACT2FN[config.hidden_act]

    def forward(self, hidden_states):
        down_proj = self.down_proj(self.act_fn(self.gate_proj(hidden_states)) * self.up_proj(hidden_states))
        return down_proj


class MiMoV2MoEGate(nn.Module):
    def __init__(self, config):
        super().__init__()
        self.config = config
        self.top_k = config.num_experts_per_tok
        self.n_routed_experts = config.n_routed_experts
        self.routed_scaling_factor = (
            config.routed_scaling_factor
            if config.routed_scaling_factor is not None
            else 1.0
        )
        self.scoring_func = config.scoring_func
        self.topk_method = config.topk_method
        self.n_group = config.n_group
        self.topk_group = config.topk_group

        # topk selection algorithm
        self.norm_topk_prob = config.norm_topk_prob
        self.gating_dim = config.hidden_size
        self.weight = nn.Parameter(
            torch.empty((self.n_routed_experts, self.gating_dim))
        )
        if self.topk_method == "noaux_tc":
            self.e_score_correction_bias = nn.Parameter(
                torch.empty((self.n_routed_experts))
            )

    def forward(self, hidden_states):
        bsz, seq_len, h = hidden_states.shape
        ### compute gating score
        hidden_states = hidden_states.view(-1, h)
        logits = F.linear(
            hidden_states.type(torch.float32), self.weight.type(torch.float32), None
        )
        if self.scoring_func == "sigmoid":
            scores = logits.sigmoid()
        else:
            raise NotImplementedError(
                f"insupportable scoring function for MoE gating: {self.scoring_func}"
            )

        ### select top-k experts
        if self.topk_method == "noaux_tc":
            assert not self.training
            scores_for_choice = scores.view(bsz * seq_len, -1) + self.e_score_correction_bias.unsqueeze(0)
            group_scores = (
                scores_for_choice.view(bsz * seq_len, self.n_group, -1).topk(2, dim=-1)[0].sum(dim = -1)
            )  # [n, n_group]
            group_idx = torch.topk(
                group_scores, k=self.topk_group, dim=-1, sorted=False
            )[
                1
            ]  # [n, top_k_group]
            group_mask = torch.zeros_like(group_scores)  # [n, n_group]
            group_mask.scatter_(1, group_idx, 1)  # [n, n_group]
            score_mask = (
                group_mask.unsqueeze(-1)
                .expand(
                    bsz * seq_len, self.n_group, self.n_routed_experts // self.n_group
                )
                .reshape(bsz * seq_len, -1)
            )  # [n, e]
            tmp_scores = scores_for_choice.masked_fill(~score_mask.bool(), float("-inf"))  # [n, e]
            _, topk_idx = torch.topk(
                tmp_scores, k=self.top_k, dim=-1, sorted=False
            )
            topk_weight = scores.gather(1, topk_idx)
        else:
            raise NotImplementedError(
                f"insupportable TopK function for MoE gating: {self.topk_method}"
            )

        ### norm gate to sum 1
        if self.top_k > 1 and self.norm_topk_prob:
            denominator = topk_weight.sum(dim=-1, keepdim=True) + 1e-20
            topk_weight = topk_weight / denominator
        topk_weight = topk_weight * self.routed_scaling_factor # must multiply the scaling factor

        return topk_idx, topk_weight


class MiMoV2MoE(nn.Module):
    """
    A mixed expert module containing shared experts.
    """

    def __init__(self, config):
        super().__init__()
        self.config = config
        self.experts = nn.ModuleList(
            [
                MiMoV2MLP(config, intermediate_size=config.moe_intermediate_size)
                for _ in range(config.n_routed_experts)
            ]
        )
        self.gate = MiMoV2MoEGate(config)

    def moe(self, hidden_states: torch.Tensor, topk_indices: torch.Tensor, topk_weights: torch.Tensor):
        r"""
        CALL FOR CONTRIBUTION! I don't have time to optimise this right now, but expert weights need to be fused
        to not have to do a loop here (deepseek has 256 experts soooo yeah).
        """
        final_hidden_states = torch.zeros_like(hidden_states, dtype=topk_weights.dtype)
        expert_mask = torch.nn.functional.one_hot(topk_indices, num_classes=len(self.experts))
        expert_mask = expert_mask.permute(2, 0, 1)

        for expert_idx in range(len(self.experts)):
            expert = self.experts[expert_idx]
            mask = expert_mask[expert_idx]
            token_indices, weight_indices = torch.where(mask)

            if token_indices.numel() > 0:
                expert_weights = topk_weights[token_indices, weight_indices]
                expert_input = hidden_states[token_indices]
                expert_output = expert(expert_input)
                weighted_output = expert_output * expert_weights.unsqueeze(-1)
                final_hidden_states.index_add_(0, token_indices, weighted_output)

        # in original deepseek, the output of the experts are gathered once we leave this module
        # thus the moe module is itelsf an IsolatedParallel module
        # and all expert are "local" meaning we shard but we don't gather
        return final_hidden_states.type(hidden_states.dtype)
    

    def forward(self, hidden_states: torch.Tensor) -> tuple[torch.Tensor, torch.Tensor]:
        orig_shape = hidden_states.shape
        topk_indices, topk_weights = self.gate(hidden_states)
        hidden_states = hidden_states.view(-1, hidden_states.shape[-1])
        hidden_states = self.moe(hidden_states, topk_indices, topk_weights).view(*orig_shape)

        return hidden_states


class MiMoV2Attention(nn.Module):
    """MiMoV2 Global Attention (pattern == 0) and Sliding Window Attention (pattern == 1)."""

    def __init__(self, config: MiMoV2FlashConfig, is_swa: bool, layer_idx: int):
        super().__init__()
        self.config = config
        self.layer_idx = layer_idx

        if is_swa:
            self.head_dim = config.swa_head_dim
            self.v_head_dim = config.swa_v_head_dim
            self.num_attention_heads = config.swa_num_attention_heads
            self.num_key_value_heads = config.swa_num_key_value_heads
        else:
            self.head_dim = config.head_dim
            self.v_head_dim = config.v_head_dim
            self.num_attention_heads = config.num_attention_heads
            self.num_key_value_heads = config.num_key_value_heads

        self.rope_dim = int(self.head_dim * config.partial_rotary_factor)
        self.num_key_value_groups = self.num_attention_heads // self.num_key_value_heads
        self.attention_bias = config.attention_bias
        self.attention_dropout: float = config.attention_dropout
        self.scaling = self.head_dim ** -0.5

        # These dimensions are for the attention layers
        q_hidden_size = self.num_attention_heads * self.head_dim
        k_hidden_size = self.num_key_value_heads * self.head_dim
        v_hidden_size = self.num_key_value_heads * self.v_head_dim
        o_hidden_size = self.num_attention_heads * self.v_head_dim

        self.q_proj = nn.Linear(config.hidden_size, q_hidden_size, bias=self.attention_bias)
        self.k_proj = nn.Linear(config.hidden_size, k_hidden_size, bias=self.attention_bias)
        self.v_proj = nn.Linear(config.hidden_size, v_hidden_size, bias=self.attention_bias)
        self.o_proj = nn.Linear(o_hidden_size, config.hidden_size, bias=False)

        self.attention_sink_bias = (
            torch.nn.Parameter(torch.empty(config.num_attention_heads), requires_grad=False)
            if (config.add_full_attention_sink_bias and not is_swa) or (config.add_swa_attention_sink_bias and is_swa)
            else None
        )

    def forward(
        self,
        hidden_states: torch.Tensor,
        position_embeddings: tuple[torch.Tensor, torch.Tensor],
        attention_mask: Optional[torch.Tensor],
        past_key_values: Optional[Cache] = None,
        cache_position: Optional[torch.LongTensor] = None,
        position_ids: Optional[torch.LongTensor] = None,
        **kwargs: Unpack[TransformersKwargs],
    ) -> tuple[torch.Tensor, torch.Tensor]:
        input_shape = hidden_states.shape[:-1]
        qk_hidden_shape = (*input_shape, -1, self.head_dim)
        v_hidden_shape = (*input_shape, -1, self.v_head_dim)

        query_states = self.q_proj(hidden_states).view(qk_hidden_shape).transpose(1, 2)
        key_states = self.k_proj(hidden_states).view(qk_hidden_shape).transpose(1, 2)
        value_states = self.v_proj(hidden_states).view(v_hidden_shape).transpose(1, 2)

        cos, sin = position_embeddings

        query_rope, query_nope = query_states.split([self.rope_dim, self.head_dim - self.rope_dim], dim=-1)
        key_rope, key_nope = key_states.split([self.rope_dim, self.head_dim - self.rope_dim], dim=-1)

        query_rope, key_rope = apply_rotary_pos_emb(query_rope, key_rope, cos, sin)

        query_states = torch.cat([query_rope, query_nope], dim=-1)
        key_states = torch.cat([key_rope, key_nope], dim=-1)

        if past_key_values is not None:
            # sin and cos are specific to RoPE models; cache_position needed for the static cache
            cache_kwargs = {"sin": sin, "cos": cos, "cache_position": cache_position}
            key_states, value_states = past_key_values.update(key_states, value_states, self.layer_idx, cache_kwargs)

        attention_interface: Callable = eager_attention_forward
        if self.config._attn_implementation != "eager":
            attention_interface = ALL_ATTENTION_FUNCTIONS[self.config._attn_implementation]

        attn_output, attn_weights = attention_interface(
            self,
            query_states,
            key_states,
            value_states,
            attention_mask,
            dropout=0.0 if not self.training else self.attention_dropout,
            scaling=self.scaling,
            position_ids=position_ids,
            sinks=self.attention_sink_bias,
        )

        attn_output = attn_output.reshape(*input_shape, -1).contiguous()
        attn_output = self.o_proj(attn_output)
        return attn_output, attn_weights


class MiMoV2DecoderLayer(nn.Module):
    """
    MiMoV2 Decoder Layer. It dynamically chooses the correct attention
    module based on the layer index and the `hybrid_layer_pattern`.
    """

    def __init__(self, config: MiMoV2FlashConfig, layer_idx: int):
        super().__init__()

        # This is the key logic: choose the module based on the pattern
        is_swa_layer = config.hybrid_layer_pattern[layer_idx] == 1
        if is_swa_layer:
            self.attention_type = "sliding_window_attention"
            self.self_attn = MiMoV2Attention(config, True, layer_idx)
        else:
            self.attention_type = "full_attention"
            self.self_attn = MiMoV2Attention(config, False, layer_idx)

        self.mlp = (
            MiMoV2MoE(config)
            if (
                    getattr(config, 'n_routed_experts', None) is not None
                    and config.moe_layer_freq[layer_idx]
            )
            else MiMoV2MLP(config)
        )

        self.input_layernorm = MiMoV2RMSNorm(config.hidden_size, eps=config.layernorm_epsilon)
        self.post_attention_layernorm = MiMoV2RMSNorm(config.hidden_size, eps=config.layernorm_epsilon)
        self.hidden_size = config.hidden_size

    def forward(
        self,
        hidden_states: torch.Tensor,
        attention_mask: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.LongTensor] = None,
        past_key_values: Optional[Cache] = None,
        use_cache: Optional[bool] = False,
        cache_position: Optional[torch.LongTensor] = None,
        position_embeddings: Optional[tuple[torch.Tensor, torch.Tensor]] = None,
        **kwargs: Unpack[TransformersKwargs],
    ) -> torch.Tensor:
        residual = hidden_states
        hidden_states = self.input_layernorm(hidden_states)
        # Self Attention
        hidden_states, _ = self.self_attn(
            hidden_states=hidden_states,
            attention_mask=attention_mask,
            position_ids=position_ids,
            past_key_values=past_key_values,
            use_cache=use_cache,
            cache_position=cache_position,
            position_embeddings=position_embeddings,
            **kwargs,
        )
        hidden_states = residual + hidden_states

        # MLP or MOE
        residual = hidden_states
        hidden_states = self.post_attention_layernorm(hidden_states)
        hidden_states = self.mlp(hidden_states)
        hidden_states = residual + hidden_states
        return hidden_states

class MiMoV2FlashRotaryEmbedding(nn.Module):
    inv_freq: torch.Tensor  # fix linting for `register_buffer`

    def __init__(self, config: MiMoV2FlashConfig, is_swa, device=None):
        super().__init__()
        # BC: "rope_type" was originally "type"
        if hasattr(config, "rope_scaling") and isinstance(config.rope_scaling, dict):
            self.rope_type = config.rope_scaling.get("rope_type", config.rope_scaling.get("type"))
        else:
            self.rope_type = "default"
        self.max_seq_len_cached = config.max_position_embeddings
        self.original_max_seq_len = config.max_position_embeddings

        self.config = config
        self.rope_init_fn = ROPE_INIT_FUNCTIONS[self.rope_type]

        if is_swa:
            self.config.rope_theta = config.swa_rope_theta
            self.config.head_dim = config.swa_head_dim

        inv_freq, self.attention_scaling = self.rope_init_fn(self.config, device)
        self.register_buffer("inv_freq", inv_freq, persistent=False)
        self.original_inv_freq = self.inv_freq

    @torch.no_grad()
    @dynamic_rope_update  # power user: used with advanced RoPE types (e.g. dynamic rope)
    def forward(self, x, position_ids):
        inv_freq_expanded = self.inv_freq[None, :, None].float().expand(position_ids.shape[0], -1, 1).to(x.device)
        position_ids_expanded = position_ids[:, None, :].float()

        device_type = x.device.type if isinstance(x.device.type, str) and x.device.type != "mps" else "cpu"
        with torch.autocast(device_type=device_type, enabled=False):  # Force float32
            freqs = (inv_freq_expanded.float() @ position_ids_expanded.float()).transpose(1, 2)
            emb = torch.cat((freqs, freqs), dim=-1)
            cos = emb.cos() * self.attention_scaling
            sin = emb.sin() * self.attention_scaling

        return cos.to(dtype=x.dtype), sin.to(dtype=x.dtype)


@auto_docstring
class MiMoV2Model(PreTrainedModel):
    """The main 'model' block, corresponding to `model.` in the weight map."""
    config_class = MiMoV2FlashConfig

    def __init__(self, config: MiMoV2FlashConfig):
        super().__init__(config)
        self.vocab_size = config.vocab_size

        self.embed_tokens = nn.Embedding(config.vocab_size, config.hidden_size)
        self.layers = nn.ModuleList(
            [MiMoV2DecoderLayer(config, layer_idx) for layer_idx in range(config.num_hidden_layers)]
        )
        self.norm = MiMoV2RMSNorm(config.hidden_size, eps=config.layernorm_epsilon)
        self.rotary_emb = MiMoV2FlashRotaryEmbedding(config=config, is_swa=False)
        self.swa_rotary_emb = MiMoV2FlashRotaryEmbedding(config=config, is_swa=True)

        self.has_sliding_layers = any(
            pattern == 1 for pattern in config.hybrid_layer_pattern
        )

        # For Huggingface DynamicCache compatibility
        self.config.layer_types = [
            "sliding_attention" if config.hybrid_layer_pattern[i] == 1 else "full_attention"
            for i in range(config.num_hidden_layers)
        ]

    @auto_docstring
    def forward(
        self,
        input_ids: Optional[torch.LongTensor] = None,
        attention_mask: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.LongTensor] = None,
        past_key_values: Optional[Cache] = None,
        inputs_embeds: Optional[torch.FloatTensor] = None,
        use_cache: Optional[bool] = None,
        cache_position: Optional[torch.LongTensor] = None,
        **kwargs: Unpack[TransformersKwargs],
    ) -> MoeModelOutputWithPast:
        if (input_ids is None) ^ (inputs_embeds is not None):
            raise ValueError("You must specify exactly one of input_ids or inputs_embeds")

        if inputs_embeds is None:
            inputs_embeds = self.embed_tokens(input_ids)

        if use_cache and past_key_values is None:
            past_key_values = DynamicCache(config=self.config)

        if cache_position is None:
            past_seen_tokens = past_key_values.get_seq_length() if past_key_values is not None else 0
            cache_position = torch.arange(
                past_seen_tokens, past_seen_tokens + inputs_embeds.shape[1], device=inputs_embeds.device
            )

        if position_ids is None:
            position_ids = cache_position.unsqueeze(0)

        # It may already have been prepared by e.g. `generate`
        if not isinstance(causal_mask_mapping := attention_mask, dict):
            # Prepare mask arguments
            mask_kwargs = {
                "config": self.config,
                "input_embeds": inputs_embeds,
                "attention_mask": attention_mask,
                "cache_position": cache_position,
                "past_key_values": past_key_values,
                "position_ids": position_ids,
            }
            # Create the masks
            causal_mask_mapping = {
                "full_attention": create_causal_mask(**mask_kwargs),
            }
            # The sliding window alternating layers are not always activated depending on the config
            if self.has_sliding_layers:
                causal_mask_mapping["sliding_window_attention"] = create_sliding_window_causal_mask(**mask_kwargs)

        hidden_states = inputs_embeds
        position_embeddings = self.rotary_emb(hidden_states, position_ids)
        swa_position_embeddings = self.swa_rotary_emb(hidden_states, position_ids)

        for decoder_layer in self.layers[: self.config.num_hidden_layers]:
            hidden_states = decoder_layer(
                hidden_states,
                attention_mask=causal_mask_mapping[decoder_layer.attention_type],
                position_embeddings=(
                    position_embeddings 
                    if decoder_layer.attention_type == "full_attention" 
                    else swa_position_embeddings
                ),
                position_ids=position_ids,
                past_key_values=past_key_values,
                use_cache=use_cache,
                cache_position=cache_position,
                **kwargs,
            )

        hidden_states = self.norm(hidden_states)
        return BaseModelOutputWithPast(
            last_hidden_state=hidden_states,
            past_key_values=past_key_values if use_cache else None,
        )


@auto_docstring
class MiMoV2FlashForCausalLM(PreTrainedModel,GenerationMixin):
    _tied_weights_keys = {"lm_head.weight": "model.embed_tokens.weight"}
    _tp_plan = {"lm_head": "colwise_rep"}
    _pp_plan = {"lm_head": (["hidden_states"], ["logits"])}

    config_class = MiMoV2FlashConfig
    _keys_to_ignore_on_load_unexpected = [r"model.layers\.\d+\.self_attn\.rotary_emb\.inv_freq"]

    def __init__(self, config: MiMoV2FlashConfig):
        super().__init__(config)
        self.model = MiMoV2Model(config)
        self.vocab_size = config.vocab_size
        self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)

        # Initialize weights and apply final processing
        self.post_init()

    @can_return_tuple
    @auto_docstring
    def forward(
        self,
        input_ids: Optional[torch.LongTensor] = None,
        attention_mask: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.LongTensor] = None,
        past_key_values: Optional[Cache] = None,
        inputs_embeds: Optional[torch.FloatTensor] = None,
        labels: Optional[torch.LongTensor] = None,
        use_cache: Optional[bool] = None,
        cache_position: Optional[torch.LongTensor] = None,
        logits_to_keep: Union[int, torch.Tensor] = 0,
        **kwargs: Unpack[TransformersKwargs],
    ) -> CausalLMOutputWithPast:
        
        outputs: BaseModelOutputWithPast = self.model(
            input_ids=input_ids,
            attention_mask=attention_mask,
            position_ids=position_ids,
            past_key_values=past_key_values,
            inputs_embeds=inputs_embeds,
            use_cache=use_cache,
            cache_position=cache_position,
            **kwargs,
        )

        hidden_states = outputs.last_hidden_state
        # Only compute necessary logits, and do not upcast them to float if we are not computing the loss
        slice_indices = slice(-logits_to_keep, None) if isinstance(logits_to_keep, int) else logits_to_keep
        logits = self.lm_head(hidden_states[:, slice_indices, :])

        loss = None
        if labels is not None:
            loss = self.loss_function(logits=logits, labels=labels, vocab_size=self.config.vocab_size, **kwargs)

        return CausalLMOutputWithPast(
            loss=loss,
            logits=logits,
            past_key_values=outputs.past_key_values,
            hidden_states=outputs.hidden_states,
            attentions=outputs.attentions,
        )

__all__ = [
    "MiMoV2FlashForCausalLM"
]