modeling_scatterbrain_moonlight.py

# coding=utf-8
# Copy from https://huggingface.co/deepseek-ai/DeepSeek-V3/blob/main/modeling_deepseek.py
# Modified to use ScatterMoE for optimized expert computation
# Copyright 2023 DeepSeek-AI 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.
""" PyTorch DeepSeek model with ScatterMoE optimization."""
import math
import warnings
from typing import Any, Dict, List, Optional, Tuple, Union

import torch
import torch.nn.functional as F
import torch.utils.checkpoint
from torch import nn
from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss

from transformers.activations import ACT2FN
from transformers.cache_utils import Cache, DynamicCache
from transformers.modeling_attn_mask_utils import (
    AttentionMaskConverter,
    _prepare_4d_attention_mask,
    _prepare_4d_causal_attention_mask,
)
from transformers.modeling_outputs import (
    BaseModelOutputWithPast,
    CausalLMOutputWithPast,
    SequenceClassifierOutputWithPast,
)
from transformers.modeling_utils import PreTrainedModel
from transformers.pytorch_utils import (
    ALL_LAYERNORM_LAYERS,
    is_torch_greater_or_equal_than_1_13,
)
from transformers.utils import (
    add_start_docstrings,
    add_start_docstrings_to_model_forward,
    is_flash_attn_2_available,
    is_flash_attn_greater_or_equal_2_10,
    logging,
    replace_return_docstrings,
)
from transformers.utils.import_utils import is_torch_fx_available
from .configuration_scatterbrain_moonlight import ScatterbrainMoonlightConfig
import torch.distributed as dist
import numpy as np

# Import ScatterMoE
try:
    from scattermoe import flatten_sort_count, parallel_linear
    SCATTERMOE_AVAILABLE = True
except ImportError:
    SCATTERMOE_AVAILABLE = False
    warnings.warn("ScatterMoE not available. Install with: pip install scattermoe")

if is_flash_attn_2_available():
    from flash_attn import flash_attn_func, flash_attn_varlen_func
    from flash_attn.bert_padding import index_first_axis, pad_input, unpad_input  # noqa


# This makes `_prepare_4d_causal_attention_mask` a leaf function in the FX graph.
if is_torch_fx_available():
    if not is_torch_greater_or_equal_than_1_13:
        import torch.fx

    _prepare_4d_causal_attention_mask = torch.fx.wrap(_prepare_4d_causal_attention_mask)


logger = logging.get_logger(__name__)

_CONFIG_FOR_DOC = "ScatterbrainMoonlightConfig"


def _get_cache_length(cache, seq_length, layer_idx=None):
    """Helper to get cache length, compatible with both old and new transformers API."""
    if hasattr(cache, 'get_usable_length'):
        if layer_idx is not None:
            return cache.get_usable_length(seq_length, layer_idx)
        return cache.get_usable_length(seq_length)
    elif hasattr(cache, 'get_seq_length'):
        if layer_idx is not None:
            return cache.get_seq_length(layer_idx)
        return cache.get_seq_length()
    return 0


class ExpandedDynamicCache(Cache):
    """Dynamic cache that supports arbitrary layer indices for virtual layers."""

    def __init__(self, num_hidden_layers: int = None):
        self._num_hidden_layers = num_hidden_layers or 128
        self.key_cache: List[Optional[torch.Tensor]] = []
        self.value_cache: List[Optional[torch.Tensor]] = []
        self._seen_tokens = 0
        self.layers = None
        self.layer_class_to_replicate = None
        self.offloading = False

    def update(
        self,
        key_states: torch.Tensor,
        value_states: torch.Tensor,
        layer_idx: int,
        cache_kwargs: Optional[dict] = None,
    ) -> Tuple[torch.Tensor, torch.Tensor]:
        while len(self.key_cache) <= layer_idx:
            self.key_cache.append(None)
            self.value_cache.append(None)

        if self.key_cache[layer_idx] is None:
            self.key_cache[layer_idx] = key_states
            self.value_cache[layer_idx] = value_states
            if layer_idx == 0:
                self._seen_tokens += key_states.shape[-2]
        else:
            self.key_cache[layer_idx] = torch.cat([self.key_cache[layer_idx], key_states], dim=-2)
            self.value_cache[layer_idx] = torch.cat([self.value_cache[layer_idx], value_states], dim=-2)
            if layer_idx == 0:
                self._seen_tokens += key_states.shape[-2]

        return self.key_cache[layer_idx], self.value_cache[layer_idx]

    def get_seq_length(self, layer_idx: int = 0) -> int:
        if layer_idx < len(self.key_cache) and self.key_cache[layer_idx] is not None:
            return self.key_cache[layer_idx].shape[-2]
        return 0

    def get_max_cache_shape(self) -> Optional[int]:
        return None

    def get_max_length(self) -> Optional[int]:
        return None

    @property
    def seen_tokens(self) -> int:
        return self._seen_tokens

    def __len__(self) -> int:
        return len(self.key_cache)

    def __iter__(self):
        for layer_idx in range(len(self)):
            if layer_idx < len(self.key_cache) and self.key_cache[layer_idx] is not None:
                yield (self.key_cache[layer_idx], self.value_cache[layer_idx])
            else:
                yield (None, None)

    def __getitem__(self, layer_idx: int):
        if layer_idx < len(self.key_cache) and self.key_cache[layer_idx] is not None:
            return (self.key_cache[layer_idx], self.value_cache[layer_idx])
        return (None, None)

    def to_legacy_cache(self):
        legacy_cache = ()
        for layer_idx in range(len(self)):
            if layer_idx < len(self.key_cache) and self.key_cache[layer_idx] is not None:
                legacy_cache += ((self.key_cache[layer_idx], self.value_cache[layer_idx]),)
            else:
                legacy_cache += ((None, None),)
        return legacy_cache

    def reorder_cache(self, beam_idx: torch.LongTensor):
        for layer_idx in range(len(self.key_cache)):
            if self.key_cache[layer_idx] is not None:
                device = self.key_cache[layer_idx].device
                self.key_cache[layer_idx] = self.key_cache[layer_idx].index_select(0, beam_idx.to(device))
                self.value_cache[layer_idx] = self.value_cache[layer_idx].index_select(0, beam_idx.to(device))


def _get_unpad_data(attention_mask):
    seqlens_in_batch = attention_mask.sum(dim=-1, dtype=torch.int32)
    indices = torch.nonzero(attention_mask.flatten(), as_tuple=False).flatten()
    max_seqlen_in_batch = seqlens_in_batch.max().item()
    cu_seqlens = F.pad(
        torch.cumsum(seqlens_in_batch, dim=0, dtype=torch.torch.int32), (1, 0)
    )
    return (
        indices,
        cu_seqlens,
        max_seqlen_in_batch,
    )


class ScatterbrainMoonlightRMSNorm(nn.Module):
    def __init__(self, hidden_size, eps=1e-6):
        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)


ALL_LAYERNORM_LAYERS.append(ScatterbrainMoonlightRMSNorm)


class ScatterbrainMoonlightRotaryEmbedding(nn.Module):
    def __init__(self, dim, max_position_embeddings=2048, base=10000, device=None):
        super().__init__()

        self.dim = dim
        self.max_position_embeddings = max_position_embeddings
        self.base = base
        inv_freq = 1.0 / (
            self.base ** (torch.arange(0, self.dim, 2).float().to(device) / self.dim)
        )
        self.register_buffer("inv_freq", inv_freq, persistent=False)

        self._set_cos_sin_cache(
            seq_len=max_position_embeddings,
            device=self.inv_freq.device,
            dtype=torch.get_default_dtype(),
        )
        self.max_seq_len_cached = None

    def _set_cos_sin_cache(self, seq_len, device, dtype):
        self.max_seq_len_cached = seq_len
        t = torch.arange(
            self.max_seq_len_cached, device=device, dtype=self.inv_freq.dtype
        )

        freqs = torch.outer(t, self.inv_freq.to(t.device))
        emb = torch.cat((freqs, freqs), dim=-1)
        self.register_buffer("cos_cached", emb.cos().to(dtype), persistent=False)
        self.register_buffer("sin_cached", emb.sin().to(dtype), persistent=False)

    def forward(self, x, seq_len=None):
        if self.max_seq_len_cached is None or seq_len > self.max_seq_len_cached:
            self._set_cos_sin_cache(seq_len=seq_len, device=x.device, dtype=x.dtype)

        return (
            self.cos_cached[:seq_len].to(dtype=x.dtype),
            self.sin_cached[:seq_len].to(dtype=x.dtype),
        )


class ScatterbrainMoonlightLinearScalingRotaryEmbedding(ScatterbrainMoonlightRotaryEmbedding):

    def __init__(
        self,
        dim,
        max_position_embeddings=2048,
        base=10000,
        device=None,
        scaling_factor=1.0,
    ):
        self.scaling_factor = scaling_factor
        super().__init__(dim, max_position_embeddings, base, device)

    def _set_cos_sin_cache(self, seq_len, device, dtype):
        self.max_seq_len_cached = seq_len
        t = torch.arange(
            self.max_seq_len_cached, device=device, dtype=self.inv_freq.dtype
        )
        t = t / self.scaling_factor

        freqs = torch.outer(t, self.inv_freq)
        emb = torch.cat((freqs, freqs), dim=-1)
        self.register_buffer("cos_cached", emb.cos().to(dtype), persistent=False)
        self.register_buffer("sin_cached", emb.sin().to(dtype), persistent=False)


class ScatterbrainMoonlightDynamicNTKScalingRotaryEmbedding(ScatterbrainMoonlightRotaryEmbedding):

    def __init__(
        self,
        dim,
        max_position_embeddings=2048,
        base=10000,
        device=None,
        scaling_factor=1.0,
    ):
        self.scaling_factor = scaling_factor
        super().__init__(dim, max_position_embeddings, base, device)

    def _set_cos_sin_cache(self, seq_len, device, dtype):
        self.max_seq_len_cached = seq_len

        if seq_len > self.max_position_embeddings:
            base = self.base * (
                (self.scaling_factor * seq_len / self.max_position_embeddings)
                - (self.scaling_factor - 1)
            ) ** (self.dim / (self.dim - 2))
            inv_freq = 1.0 / (
                base ** (torch.arange(0, self.dim, 2).float().to(device) / self.dim)
            )
            self.register_buffer("inv_freq", inv_freq, persistent=False)

        t = torch.arange(
            self.max_seq_len_cached, device=device, dtype=self.inv_freq.dtype
        )

        freqs = torch.outer(t, self.inv_freq)
        emb = torch.cat((freqs, freqs), dim=-1)
        self.register_buffer("cos_cached", emb.cos().to(dtype), persistent=False)
        self.register_buffer("sin_cached", emb.sin().to(dtype), persistent=False)


def yarn_find_correction_dim(
    num_rotations, dim, base=10000, max_position_embeddings=2048
):
    return (dim * math.log(max_position_embeddings / (num_rotations * 2 * math.pi))) / (
        2 * math.log(base)
    )


def yarn_find_correction_range(
    low_rot, high_rot, dim, base=10000, max_position_embeddings=2048
):
    low = math.floor(
        yarn_find_correction_dim(low_rot, dim, base, max_position_embeddings)
    )
    high = math.ceil(
        yarn_find_correction_dim(high_rot, dim, base, max_position_embeddings)
    )
    return max(low, 0), min(high, dim - 1)


def yarn_get_mscale(scale=1, mscale=1):
    if scale <= 1:
        return 1.0
    return 0.1 * mscale * math.log(scale) + 1.0


def yarn_linear_ramp_mask(min, max, dim):
    if min == max:
        max += 0.001

    linear_func = (torch.arange(dim, dtype=torch.float32) - min) / (max - min)
    ramp_func = torch.clamp(linear_func, 0, 1)
    return ramp_func


class ScatterbrainMoonlightYarnRotaryEmbedding(ScatterbrainMoonlightRotaryEmbedding):

    def __init__(
        self,
        dim,
        max_position_embeddings=2048,
        base=10000,
        device=None,
        scaling_factor=1.0,
        original_max_position_embeddings=4096,
        beta_fast=32,
        beta_slow=1,
        mscale=1,
        mscale_all_dim=0,
    ):
        self.scaling_factor = scaling_factor
        self.original_max_position_embeddings = original_max_position_embeddings
        self.beta_fast = beta_fast
        self.beta_slow = beta_slow
        self.mscale = mscale
        self.mscale_all_dim = mscale_all_dim
        super().__init__(dim, max_position_embeddings, base, device)

    def _set_cos_sin_cache(self, seq_len, device, dtype):
        self.max_seq_len_cached = seq_len
        dim = self.dim

        freq_extra = 1.0 / (
            self.base
            ** (torch.arange(0, dim, 2, dtype=torch.float32, device=device) / dim)
        )
        freq_inter = 1.0 / (
            self.scaling_factor
            * self.base
            ** (torch.arange(0, dim, 2, dtype=torch.float32, device=device) / dim)
        )

        low, high = yarn_find_correction_range(
            self.beta_fast,
            self.beta_slow,
            dim,
            self.base,
            self.original_max_position_embeddings,
        )
        inv_freq_mask = 1.0 - yarn_linear_ramp_mask(low, high, dim // 2).to(
            device=device, dtype=torch.float32
        )
        inv_freq = freq_inter * (1 - inv_freq_mask) + freq_extra * inv_freq_mask
        self.register_buffer("inv_freq", inv_freq, persistent=False)

        t = torch.arange(seq_len, device=device, dtype=torch.float32)

        freqs = torch.outer(t, inv_freq)

        _mscale = float(
            yarn_get_mscale(self.scaling_factor, self.mscale)
            / yarn_get_mscale(self.scaling_factor, self.mscale_all_dim)
        )

        emb = torch.cat((freqs, freqs), dim=-1)
        self.register_buffer(
            "cos_cached", (emb.cos() * _mscale).to(dtype), persistent=False
        )
        self.register_buffer(
            "sin_cached", (emb.sin() * _mscale).to(dtype), persistent=False
        )


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, unsqueeze_dim=1):
    cos = cos[position_ids].unsqueeze(unsqueeze_dim)
    sin = sin[position_ids].unsqueeze(unsqueeze_dim)

    b, h, s, d = q.shape
    q = q.view(b, h, s, d // 2, 2).transpose(4, 3).reshape(b, h, s, d)

    b, h, s, d = k.shape
    k = k.view(b, h, s, d // 2, 2).transpose(4, 3).reshape(b, h, s, d)

    q_embed = (q * cos) + (rotate_half(q) * sin)
    k_embed = (k * cos) + (rotate_half(k) * sin)
    return q_embed, k_embed


class ScatterbrainMoonlightMLP(nn.Module):
    def __init__(self, config, hidden_size=None, intermediate_size=None):
        super().__init__()
        self.config = config
        self.hidden_size = config.hidden_size if hidden_size is None else 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, x):
        down_proj = self.down_proj(self.act_fn(self.gate_proj(x)) * self.up_proj(x))
        return down_proj


class ScatterbrainMoonlightMoEGate(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
        self.scoring_func = config.scoring_func
        self.seq_aux = config.seq_aux
        self.topk_method = config.topk_method
        self.n_group = config.n_group
        self.topk_group = config.topk_group

        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))
            )
        self.reset_parameters()

    def reset_parameters(self) -> None:
        import torch.nn.init as init

        init.kaiming_uniform_(self.weight, a=math.sqrt(5))

    def forward(self, hidden_states):
        bsz, seq_len, h = hidden_states.shape
        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}"
            )

        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)
            )
            group_idx = torch.topk(
                group_scores, k=self.topk_group, dim=-1, sorted=False
            )[1]
            group_mask = torch.zeros_like(group_scores)
            group_mask.scatter_(1, group_idx, 1)
            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)
            )
            tmp_scores = scores_for_choice.masked_fill(~score_mask.bool(), 0.0)
            _, 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}"
            )

        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

        return topk_idx, topk_weight


class ScatterbrainMoonlightMoE(nn.Module):
    """
    A mixed expert module using ScatterMoE for optimized expert computation.

    Uses stacked expert weights and ScatterMoE's parallel_linear for efficient
    GPU computation instead of nn.ModuleList with sequential expert calls.
    """

    def __init__(self, config):
        super().__init__()
        self.config = config
        self.num_experts_per_tok = config.num_experts_per_tok
        self.n_routed_experts = config.n_routed_experts
        self.hidden_size = config.hidden_size
        self.moe_intermediate_size = config.moe_intermediate_size

        # Stacked expert weights as Parameters
        # Shape: [num_experts, out_features, in_features]
        self.expert_gate_proj = nn.Parameter(
            torch.empty(config.n_routed_experts, config.moe_intermediate_size, config.hidden_size)
        )
        self.expert_up_proj = nn.Parameter(
            torch.empty(config.n_routed_experts, config.moe_intermediate_size, config.hidden_size)
        )
        self.expert_down_proj = nn.Parameter(
            torch.empty(config.n_routed_experts, config.hidden_size, config.moe_intermediate_size)
        )

        # Initialize weights
        for param in [self.expert_gate_proj, self.expert_up_proj, self.expert_down_proj]:
            nn.init.kaiming_uniform_(param, a=5**0.5)

        self.gate = ScatterbrainMoonlightMoEGate(config)

        if config.n_shared_experts is not None:
            intermediate_size = config.moe_intermediate_size * config.n_shared_experts
            self.shared_experts = ScatterbrainMoonlightMLP(
                config=config, intermediate_size=intermediate_size
            )

    def forward(self, hidden_states):
        identity = hidden_states
        orig_shape = hidden_states.shape
        topk_idx, topk_weight = self.gate(hidden_states)
        hidden_states = hidden_states.view(-1, hidden_states.shape[-1])

        if SCATTERMOE_AVAILABLE:
            y = self._forward_scattermoe(hidden_states, topk_idx, topk_weight)
        else:
            y = self._forward_loop(hidden_states, topk_idx, topk_weight)

        y = y.view(*orig_shape)

        if self.config.n_shared_experts is not None:
            y = y + self.shared_experts(identity)
        return y

    def _forward_scattermoe(
        self,
        hidden_states: torch.Tensor,
        selected_experts: torch.Tensor,
        routing_weights: torch.Tensor,
    ) -> torch.Tensor:
        """Forward pass using ScatterMoE Triton kernels."""
        sorted_expert_idxs, sorted_scattered_idxs, expert_offsets = flatten_sort_count(
            selected_experts, num_experts=self.n_routed_experts
        )

        # First pass: gate and up projections (hidden -> intermediate)
        gate_out = parallel_linear(
            inputs=hidden_states,
            expert_weights=self.expert_gate_proj.permute(0, 2, 1),
            k=self.num_experts_per_tok,
            sorted_expert_idxs=sorted_expert_idxs,
            sorted_scattered_idxs=sorted_scattered_idxs,
            expert_offsets=expert_offsets,
            grouped_out=True,
        )

        up_out = parallel_linear(
            inputs=hidden_states,
            expert_weights=self.expert_up_proj.permute(0, 2, 1),
            k=self.num_experts_per_tok,
            sorted_expert_idxs=sorted_expert_idxs,
            sorted_scattered_idxs=sorted_scattered_idxs,
            expert_offsets=expert_offsets,
            grouped_out=True,
        )

        # Apply SiLU activation (SwiGLU)
        activated = F.silu(gate_out) * up_out

        # Second pass: down projection with routing weights
        # Convert gates to match expert weight dtype (scattermoe requires consistent dtypes)
        output = parallel_linear(
            inputs=activated,
            expert_weights=self.expert_down_proj.permute(0, 2, 1),
            k=1,
            sorted_expert_idxs=sorted_expert_idxs,
            sorted_scattered_idxs=sorted_scattered_idxs,
            expert_offsets=expert_offsets,
            gates=routing_weights.to(self.expert_down_proj.dtype),
            grouped_in=True,
            grouped_out=False,
        )

        return output

    def _forward_loop(
        self,
        hidden_states: torch.Tensor,
        selected_experts: torch.Tensor,
        routing_weights: torch.Tensor,
    ) -> torch.Tensor:
        """Fallback forward pass using Python loop."""
        num_tokens = hidden_states.shape[0]

        flat_expert_indices = selected_experts.view(-1)
        flat_token_indices = torch.arange(num_tokens, device=hidden_states.device).unsqueeze(1).expand(-1, self.num_experts_per_tok).reshape(-1)
        flat_routing_weights = routing_weights.view(-1)

        sorted_indices = torch.argsort(flat_expert_indices, stable=True)
        sorted_expert_indices = flat_expert_indices[sorted_indices]
        sorted_token_indices = flat_token_indices[sorted_indices]
        sorted_routing_weights = flat_routing_weights[sorted_indices]
        sorted_hidden = hidden_states[sorted_token_indices]

        expert_counts = torch.bincount(sorted_expert_indices, minlength=self.n_routed_experts)
        expert_offsets = torch.zeros(self.n_routed_experts + 1, dtype=torch.long, device=hidden_states.device)
        expert_offsets[1:] = torch.cumsum(expert_counts, dim=0)

        final_hidden_states = torch.zeros_like(hidden_states)

        for expert_idx in range(self.n_routed_experts):
            start = expert_offsets[expert_idx].item()
            end = expert_offsets[expert_idx + 1].item()

            if start == end:
                continue

            expert_tokens = sorted_hidden[start:end]
            expert_weights = sorted_routing_weights[start:end].unsqueeze(-1)
            token_indices = sorted_token_indices[sorted_indices[start:end]]

            gate = self.expert_gate_proj[expert_idx]
            up = self.expert_up_proj[expert_idx]
            down = self.expert_down_proj[expert_idx]

            gate_out = F.linear(expert_tokens, gate)
            up_out = F.linear(expert_tokens, up)
            expert_out = F.linear(F.silu(gate_out) * up_out, down)

            final_hidden_states.index_add_(0, token_indices, expert_out * expert_weights)

        return final_hidden_states


def repeat_kv(hidden_states: torch.Tensor, n_rep: int) -> torch.Tensor:
    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)


class ScatterbrainMoonlightAttention(nn.Module):
    """Multi-headed attention from 'Attention Is All You Need' paper"""

    def __init__(self, config: ScatterbrainMoonlightConfig, layer_idx: Optional[int] = None):
        super().__init__()
        self.config = config
        self.layer_idx = layer_idx
        if layer_idx is None:
            logger.warning_once(
                f"Instantiating {self.__class__.__name__} without passing `layer_idx` is not recommended and will "
                "to errors during the forward call, if caching is used. Please make sure to provide a `layer_idx` "
                "when creating this class."
            )

        self.attention_dropout = config.attention_dropout
        self.hidden_size = config.hidden_size
        self.num_heads = config.num_attention_heads

        self.max_position_embeddings = config.max_position_embeddings
        self.rope_theta = config.rope_theta
        self.q_lora_rank = config.q_lora_rank
        self.qk_rope_head_dim = config.qk_rope_head_dim
        self.kv_lora_rank = config.kv_lora_rank
        self.v_head_dim = config.v_head_dim
        self.qk_nope_head_dim = config.qk_nope_head_dim
        self.q_head_dim = config.qk_nope_head_dim + config.qk_rope_head_dim

        self.is_causal = True

        if self.q_lora_rank is None:
            self.q_proj = nn.Linear(
                self.hidden_size, self.num_heads * self.q_head_dim, bias=False
            )
        else:
            self.q_a_proj = nn.Linear(
                self.hidden_size, config.q_lora_rank, bias=config.attention_bias
            )
            self.q_a_layernorm = ScatterbrainMoonlightRMSNorm(config.q_lora_rank)
            self.q_b_proj = nn.Linear(
                config.q_lora_rank, self.num_heads * self.q_head_dim, bias=False
            )

        self.kv_a_proj_with_mqa = nn.Linear(
            self.hidden_size,
            config.kv_lora_rank + config.qk_rope_head_dim,
            bias=config.attention_bias,
        )
        self.kv_a_layernorm = ScatterbrainMoonlightRMSNorm(config.kv_lora_rank)
        self.kv_b_proj = nn.Linear(
            config.kv_lora_rank,
            self.num_heads
            * (self.q_head_dim - self.qk_rope_head_dim + self.v_head_dim),
            bias=False,
        )

        self.o_proj = nn.Linear(
            self.num_heads * self.v_head_dim,
            self.hidden_size,
            bias=config.attention_bias,
        )
        self._init_rope()

        self.softmax_scale = self.q_head_dim ** (-0.5)
        if self.config.rope_scaling is not None:
            mscale_all_dim = self.config.rope_scaling.get("mscale_all_dim", 0)
            scaling_factor = self.config.rope_scaling["factor"]
            if mscale_all_dim:
                mscale = yarn_get_mscale(scaling_factor, mscale_all_dim)
                self.softmax_scale = self.softmax_scale * mscale * mscale

    def _init_rope(self):
        if self.config.rope_scaling is None:
            self.rotary_emb = ScatterbrainMoonlightRotaryEmbedding(
                self.qk_rope_head_dim,
                max_position_embeddings=self.max_position_embeddings,
                base=self.rope_theta,
            )
        else:
            scaling_type = self.config.rope_scaling["type"]
            scaling_factor = self.config.rope_scaling["factor"]
            if scaling_type == "linear":
                self.rotary_emb = ScatterbrainMoonlightLinearScalingRotaryEmbedding(
                    self.qk_rope_head_dim,
                    max_position_embeddings=self.max_position_embeddings,
                    scaling_factor=scaling_factor,
                    base=self.rope_theta,
                )
            elif scaling_type == "dynamic":
                self.rotary_emb = ScatterbrainMoonlightDynamicNTKScalingRotaryEmbedding(
                    self.qk_rope_head_dim,
                    max_position_embeddings=self.max_position_embeddings,
                    scaling_factor=scaling_factor,
                    base=self.rope_theta,
                )
            elif scaling_type == "yarn":
                kwargs = {
                    key: self.config.rope_scaling[key]
                    for key in [
                        "original_max_position_embeddings",
                        "beta_fast",
                        "beta_slow",
                        "mscale",
                        "mscale_all_dim",
                    ]
                    if key in self.config.rope_scaling
                }
                self.rotary_emb = ScatterbrainMoonlightYarnRotaryEmbedding(
                    self.qk_rope_head_dim,
                    max_position_embeddings=self.max_position_embeddings,
                    scaling_factor=scaling_factor,
                    base=self.rope_theta,
                    **kwargs,
                )
            else:
                raise ValueError(f"Unknown RoPE scaling type {scaling_type}")

    def _shape(self, tensor: torch.Tensor, seq_len: int, bsz: int):
        return (
            tensor.view(bsz, seq_len, self.num_heads, self.v_head_dim)
            .transpose(1, 2)
            .contiguous()
        )

    def forward(
        self,
        hidden_states: torch.Tensor,
        attention_mask: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.LongTensor] = None,
        past_key_value: Optional[Cache] = None,
        output_attentions: bool = False,
        use_cache: bool = False,
        **kwargs,
    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
        if "padding_mask" in kwargs:
            warnings.warn(
                "Passing `padding_mask` is deprecated and will be removed in v4.37. Please make sure use `attention_mask` instead.`"
            )
        bsz, q_len, _ = hidden_states.size()

        if self.q_lora_rank is None:
            q = self.q_proj(hidden_states)
        else:
            q = self.q_b_proj(self.q_a_layernorm(self.q_a_proj(hidden_states)))
        q = q.view(bsz, q_len, self.num_heads, self.q_head_dim).transpose(1, 2)
        q_nope, q_pe = torch.split(
            q, [self.qk_nope_head_dim, self.qk_rope_head_dim], dim=-1
        )

        compressed_kv = self.kv_a_proj_with_mqa(hidden_states)
        compressed_kv, k_pe = torch.split(
            compressed_kv, [self.kv_lora_rank, self.qk_rope_head_dim], dim=-1
        )
        k_pe = k_pe.view(bsz, q_len, 1, self.qk_rope_head_dim).transpose(1, 2)
        kv = (
            self.kv_b_proj(self.kv_a_layernorm(compressed_kv))
            .view(bsz, q_len, self.num_heads, self.qk_nope_head_dim + self.v_head_dim)
            .transpose(1, 2)
        )

        k_nope, value_states = torch.split(
            kv, [self.qk_nope_head_dim, self.v_head_dim], dim=-1
        )
        kv_seq_len = value_states.shape[-2]
        if past_key_value is not None:
            if self.layer_idx is None:
                raise ValueError(
                    f"The cache structure has changed since version v4.36. If you are using {self.__class__.__name__} "
                    "for auto-regressive decoding with k/v caching, please make sure to initialize the attention class "
                    "with a layer index."
                )
            kv_seq_len += _get_cache_length(past_key_value, kv_seq_len, self.layer_idx)
        cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)

        q_pe, k_pe = apply_rotary_pos_emb(q_pe, k_pe, cos, sin, position_ids)

        query_states = k_pe.new_empty(bsz, self.num_heads, q_len, self.q_head_dim)
        query_states[:, :, :, : self.qk_nope_head_dim] = q_nope
        query_states[:, :, :, self.qk_nope_head_dim :] = q_pe

        key_states = k_pe.new_empty(bsz, self.num_heads, q_len, self.q_head_dim)
        key_states[:, :, :, : self.qk_nope_head_dim] = k_nope
        key_states[:, :, :, self.qk_nope_head_dim :] = k_pe
        if past_key_value is not None:
            cache_kwargs = {"sin": sin, "cos": cos}
            key_states, value_states = past_key_value.update(
                key_states, value_states, self.layer_idx, cache_kwargs
            )

        attn_weights = (
            torch.matmul(query_states, key_states.transpose(2, 3)) * self.softmax_scale
        )

        if attn_weights.size() != (bsz, self.num_heads, q_len, kv_seq_len):
            raise ValueError(
                f"Attention weights should be of size {(bsz, self.num_heads, q_len, kv_seq_len)}, but is"
                f" {attn_weights.size()}"
            )
        if attention_mask is not None:
            if attention_mask.size() != (bsz, 1, q_len, kv_seq_len):
                raise ValueError(
                    f"Attention mask should be of size {(bsz, 1, q_len, kv_seq_len)}, but is {attention_mask.size()}"
                )
            attn_weights = attn_weights + attention_mask

        attn_weights = nn.functional.softmax(
            attn_weights, dim=-1, dtype=torch.float32
        ).to(query_states.dtype)
        attn_weights = nn.functional.dropout(
            attn_weights, p=self.attention_dropout, training=self.training
        )
        attn_output = torch.matmul(attn_weights, value_states)

        if attn_output.size() != (bsz, self.num_heads, q_len, self.v_head_dim):
            raise ValueError(
                f"`attn_output` should be of size {(bsz, self.num_heads, q_len, self.v_head_dim)}, but is"
                f" {attn_output.size()}"
            )

        attn_output = attn_output.transpose(1, 2).contiguous()

        attn_output = attn_output.reshape(bsz, q_len, self.num_heads * self.v_head_dim)

        attn_output = self.o_proj(attn_output)

        if not output_attentions:
            attn_weights = None

        return attn_output, attn_weights, past_key_value


class ScatterbrainMoonlightFlashAttention2(ScatterbrainMoonlightAttention):
    """Flash Attention 2 implementation."""

    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
        self._flash_attn_uses_top_left_mask = not is_flash_attn_greater_or_equal_2_10()

    def forward(
        self,
        hidden_states: torch.Tensor,
        attention_mask: Optional[torch.LongTensor] = None,
        position_ids: Optional[torch.LongTensor] = None,
        past_key_value: Optional[Cache] = None,
        output_attentions: bool = False,
        use_cache: bool = False,
        **kwargs,
    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
        if "padding_mask" in kwargs:
            warnings.warn(
                "Passing `padding_mask` is deprecated and will be removed in v4.37. Please make sure use `attention_mask` instead.`"
            )
            attention_mask = kwargs.pop("padding_mask")

        output_attentions = False

        bsz, q_len, _ = hidden_states.size()

        if self.q_lora_rank is None:
            q = self.q_proj(hidden_states)
        else:
            q = self.q_b_proj(self.q_a_layernorm(self.q_a_proj(hidden_states)))
        q = q.view(bsz, q_len, self.num_heads, self.q_head_dim).transpose(1, 2)
        q_nope, q_pe = torch.split(
            q, [self.qk_nope_head_dim, self.qk_rope_head_dim], dim=-1
        )

        compressed_kv = self.kv_a_proj_with_mqa(hidden_states)
        compressed_kv, k_pe = torch.split(
            compressed_kv, [self.kv_lora_rank, self.qk_rope_head_dim], dim=-1
        )
        k_pe = k_pe.view(bsz, q_len, 1, self.qk_rope_head_dim).transpose(1, 2)
        kv = (
            self.kv_b_proj(self.kv_a_layernorm(compressed_kv))
            .view(bsz, q_len, self.num_heads, self.qk_nope_head_dim + self.v_head_dim)
            .transpose(1, 2)
        )

        k_nope, value_states = torch.split(
            kv, [self.qk_nope_head_dim, self.v_head_dim], dim=-1
        )
        kv_seq_len = value_states.shape[-2]

        if past_key_value is not None:
            kv_seq_len += _get_cache_length(past_key_value, kv_seq_len, self.layer_idx)

        cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
        q_pe, k_pe = apply_rotary_pos_emb(q_pe, k_pe, cos, sin, position_ids)

        query_states = k_pe.new_empty(bsz, self.num_heads, q_len, self.q_head_dim)
        query_states[:, :, :, : self.qk_nope_head_dim] = q_nope
        query_states[:, :, :, self.qk_nope_head_dim :] = q_pe

        key_states = k_pe.new_empty(bsz, self.num_heads, q_len, self.q_head_dim)
        key_states[:, :, :, : self.qk_nope_head_dim] = k_nope
        key_states[:, :, :, self.qk_nope_head_dim :] = k_pe

        if self.q_head_dim != self.v_head_dim:
            value_states = F.pad(value_states, [0, self.q_head_dim - self.v_head_dim])

        if past_key_value is not None:
            cache_kwargs = {"sin": sin, "cos": cos}
            key_states, value_states = past_key_value.update(
                key_states, value_states, self.layer_idx, cache_kwargs
            )

        query_states = query_states.transpose(1, 2)
        key_states = key_states.transpose(1, 2)
        value_states = value_states.transpose(1, 2)

        dropout_rate = self.attention_dropout if self.training else 0.0

        input_dtype = query_states.dtype
        if input_dtype == torch.float32:
            if hasattr(self.config, "_pre_quantization_dtype"):
                target_dtype = self.config._pre_quantization_dtype
            elif torch.is_autocast_enabled():
                target_dtype = torch.get_autocast_gpu_dtype()
            else:
                target_dtype = (
                    self.q_proj.weight.dtype
                    if self.q_lora_rank is None
                    else self.q_a_proj.weight.dtype
                )

            logger.warning_once(
                f"The input hidden states seems to be silently casted in float32, this might be related to"
                f" the fact you have upcasted embedding or layer norm layers in float32. We will cast back the input in"
                f" {target_dtype}."
            )

            query_states = query_states.to(target_dtype)
            key_states = key_states.to(target_dtype)
            value_states = value_states.to(target_dtype)

        attn_output = self._flash_attention_forward(
            query_states,
            key_states,
            value_states,
            attention_mask,
            q_len,
            dropout=dropout_rate,
            softmax_scale=self.softmax_scale,
        )
        if self.q_head_dim != self.v_head_dim:
            attn_output = attn_output[:, :, :, : self.v_head_dim]

        attn_output = attn_output.reshape(
            bsz, q_len, self.num_heads * self.v_head_dim
        ).contiguous()
        attn_output = self.o_proj(attn_output)

        if not output_attentions:
            attn_weights = None

        return attn_output, attn_weights, past_key_value

    def _flash_attention_forward(
        self,
        query_states,
        key_states,
        value_states,
        attention_mask,
        query_length,
        dropout=0.0,
        softmax_scale=None,
    ):
        if not self._flash_attn_uses_top_left_mask:
            causal = self.is_causal
        else:
            causal = self.is_causal and query_length != 1

        if attention_mask is not None:
            batch_size = query_states.shape[0]
            (
                query_states,
                key_states,
                value_states,
                indices_q,
                cu_seq_lens,
                max_seq_lens,
            ) = self._upad_input(
                query_states, key_states, value_states, attention_mask, query_length
            )

            cu_seqlens_q, cu_seqlens_k = cu_seq_lens
            max_seqlen_in_batch_q, max_seqlen_in_batch_k = max_seq_lens

            attn_output_unpad = flash_attn_varlen_func(
                query_states,
                key_states,
                value_states,
                cu_seqlens_q=cu_seqlens_q,
                cu_seqlens_k=cu_seqlens_k,
                max_seqlen_q=max_seqlen_in_batch_q,
                max_seqlen_k=max_seqlen_in_batch_k,
                dropout_p=dropout,
                softmax_scale=softmax_scale,
                causal=causal,
            )

            attn_output = pad_input(
                attn_output_unpad, indices_q, batch_size, query_length
            )
        else:
            attn_output = flash_attn_func(
                query_states,
                key_states,
                value_states,
                dropout,
                softmax_scale=softmax_scale,
                causal=causal,
            )

        return attn_output

    def _upad_input(
        self, query_layer, key_layer, value_layer, attention_mask, query_length
    ):
        indices_k, cu_seqlens_k, max_seqlen_in_batch_k = _get_unpad_data(attention_mask)
        batch_size, kv_seq_len, num_key_value_heads, head_dim = key_layer.shape

        key_layer = index_first_axis(
            key_layer.reshape(batch_size * kv_seq_len, num_key_value_heads, head_dim),
            indices_k,
        )
        value_layer = index_first_axis(
            value_layer.reshape(batch_size * kv_seq_len, num_key_value_heads, head_dim),
            indices_k,
        )
        if query_length == kv_seq_len:
            query_layer = index_first_axis(
                query_layer.reshape(batch_size * kv_seq_len, self.num_heads, head_dim),
                indices_k,
            )
            cu_seqlens_q = cu_seqlens_k
            max_seqlen_in_batch_q = max_seqlen_in_batch_k
            indices_q = indices_k
        elif query_length == 1:
            max_seqlen_in_batch_q = 1
            cu_seqlens_q = torch.arange(
                batch_size + 1, dtype=torch.int32, device=query_layer.device
            )
            indices_q = cu_seqlens_q[:-1]
            query_layer = query_layer.squeeze(1)
        else:
            attention_mask = attention_mask[:, -query_length:]
            query_layer, indices_q, cu_seqlens_q, max_seqlen_in_batch_q = unpad_input(
                query_layer, attention_mask
            )

        return (
            query_layer,
            key_layer,
            value_layer,
            indices_q,
            (cu_seqlens_q, cu_seqlens_k),
            (max_seqlen_in_batch_q, max_seqlen_in_batch_k),
        )


ATTENTION_CLASSES = {
    "eager": ScatterbrainMoonlightAttention,
    "flash_attention_2": ScatterbrainMoonlightFlashAttention2,
}


class ScatterbrainMoonlightDecoderLayer(nn.Module):
    def __init__(self, config: ScatterbrainMoonlightConfig, layer_idx: int):
        super().__init__()
        self.hidden_size = config.hidden_size

        self.self_attn = ATTENTION_CLASSES[config._attn_implementation](
            config=config, layer_idx=layer_idx
        )

        self.mlp = (
            ScatterbrainMoonlightMoE(config)
            if (
                config.n_routed_experts is not None
                and layer_idx >= config.first_k_dense_replace
                and layer_idx % config.moe_layer_freq == 0
            )
            else ScatterbrainMoonlightMLP(config)
        )
        self.input_layernorm = ScatterbrainMoonlightRMSNorm(
            config.hidden_size, eps=config.rms_norm_eps
        )
        self.post_attention_layernorm = ScatterbrainMoonlightRMSNorm(
            config.hidden_size, eps=config.rms_norm_eps
        )

    def forward(
        self,
        hidden_states: torch.Tensor,
        attention_mask: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.LongTensor] = None,
        past_key_value: Optional[Tuple[torch.Tensor]] = None,
        output_attentions: Optional[bool] = False,
        use_cache: Optional[bool] = False,
        **kwargs,
    ) -> Tuple[
        torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]
    ]:
        if "padding_mask" in kwargs:
            warnings.warn(
                "Passing `padding_mask` is deprecated and will be removed in v4.37. Please make sure use `attention_mask` instead.`"
            )
        residual = hidden_states

        hidden_states = self.input_layernorm(hidden_states)

        hidden_states, self_attn_weights, present_key_value = self.self_attn(
            hidden_states=hidden_states,
            attention_mask=attention_mask,
            position_ids=position_ids,
            past_key_value=past_key_value,
            output_attentions=output_attentions,
            use_cache=use_cache,
            **kwargs,
        )
        hidden_states = residual + hidden_states

        residual = hidden_states
        hidden_states = self.post_attention_layernorm(hidden_states)
        hidden_states = self.mlp(hidden_states)
        hidden_states = residual + hidden_states

        outputs = (hidden_states,)

        if output_attentions:
            outputs += (self_attn_weights,)

        if use_cache:
            outputs += (present_key_value,)

        return outputs


DeepseekV3_START_DOCSTRING = r"""
    This model inherits from [`PreTrainedModel`]. Check the superclass documentation for the generic methods the
    library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads
    etc.)
"""


class ScatterbrainMoonlightPreTrainedModel(PreTrainedModel):
    config_class = ScatterbrainMoonlightConfig
    base_model_prefix = "model"
    supports_gradient_checkpointing = True
    _no_split_modules = ["ScatterbrainMoonlightDecoderLayer"]
    _skip_keys_device_placement = "past_key_values"
    _supports_flash_attn_2 = True
    _supports_cache_class = True

    def _init_weights(self, module):
        std = self.config.initializer_range
        if isinstance(module, nn.Linear):
            module.weight.data.normal_(mean=0.0, std=std)
            if module.bias is not None:
                module.bias.data.zero_()
        elif isinstance(module, nn.Embedding):
            module.weight.data.normal_(mean=0.0, std=std)
            if module.padding_idx is not None:
                module.weight.data[module.padding_idx].zero_()


class ScatterbrainMoonlightModel(ScatterbrainMoonlightPreTrainedModel):
    def __init__(self, config: ScatterbrainMoonlightConfig):
        super().__init__(config)
        self.padding_idx = config.pad_token_id
        self.vocab_size = config.vocab_size

        self.embed_tokens = nn.Embedding(
            config.vocab_size, config.hidden_size, self.padding_idx
        )
        self.layers = nn.ModuleList(
            [
                ScatterbrainMoonlightDecoderLayer(config, layer_idx)
                for layer_idx in range(config.num_hidden_layers)
            ]
        )
        self._attn_implementation = config._attn_implementation
        self.norm = ScatterbrainMoonlightRMSNorm(config.hidden_size, eps=config.rms_norm_eps)

        self.gradient_checkpointing = False

        self.num_loop_iterations = getattr(config, 'num_loop_iterations', 1)
        self._num_virtual_layers = 1 + self.num_loop_iterations + 1

        self.post_init()

    def get_input_embeddings(self):
        return self.embed_tokens

    def set_input_embeddings(self, value):
        self.embed_tokens = value

    def _prepare_decoder_attention_mask(
        self, attention_mask, input_shape, inputs_embeds, past_key_values_length
    ):
        combined_attention_mask = None
        if input_shape[-1] > 1:
            combined_attention_mask = _prepare_4d_causal_attention_mask(
                attention_mask,
                input_shape,
                inputs_embeds,
                past_key_values_length,
            )

        if attention_mask is not None and combined_attention_mask is not None:
            combined_attention_mask = combined_attention_mask.to(attention_mask.device)

        return combined_attention_mask

    def forward(
        self,
        input_ids: torch.LongTensor = None,
        attention_mask: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.LongTensor] = None,
        past_key_values: Optional[List[torch.FloatTensor]] = None,
        inputs_embeds: Optional[torch.FloatTensor] = None,
        use_cache: Optional[bool] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
    ) -> Union[Tuple, BaseModelOutputWithPast]:
        output_attentions = (
            output_attentions
            if output_attentions is not None
            else self.config.output_attentions
        )
        output_hidden_states = (
            output_hidden_states
            if output_hidden_states is not None
            else self.config.output_hidden_states
        )
        use_cache = use_cache if use_cache is not None else self.config.use_cache

        return_dict = (
            return_dict if return_dict is not None else self.config.use_return_dict
        )

        if input_ids is not None and inputs_embeds is not None:
            raise ValueError(
                "You cannot specify both input_ids and inputs_embeds at the same time"
            )
        elif input_ids is not None:
            batch_size, seq_length = input_ids.shape[:2]
        elif inputs_embeds is not None:
            batch_size, seq_length = inputs_embeds.shape[:2]
        else:
            raise ValueError("You have to specify either input_ids or inputs_embeds")

        if self.gradient_checkpointing and self.training:
            if use_cache:
                logger.warning_once(
                    "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                )
                use_cache = False

        past_key_values_length = 0

        if use_cache:
            if past_key_values is None:
                past_key_values = ExpandedDynamicCache(self._num_virtual_layers)
            elif not isinstance(past_key_values, ExpandedDynamicCache):
                past_key_values = ExpandedDynamicCache(self._num_virtual_layers)
            past_key_values_length = past_key_values.get_seq_length(0)

        if position_ids is None:
            device = input_ids.device if input_ids is not None else inputs_embeds.device
            position_ids = torch.arange(
                past_key_values_length,
                seq_length + past_key_values_length,
                dtype=torch.long,
                device=device,
            )
            position_ids = position_ids.unsqueeze(0)

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

        if self._attn_implementation == "flash_attention_2":
            attention_mask = (
                attention_mask
                if (attention_mask is not None and 0 in attention_mask)
                else None
            )
        else:
            attention_mask = self._prepare_decoder_attention_mask(
                attention_mask,
                (batch_size, seq_length),
                inputs_embeds,
                past_key_values_length,
            )

        hidden_states = inputs_embeds

        all_hidden_states = () if output_hidden_states else None
        all_self_attns = () if output_attentions else None
        next_decoder_cache = past_key_values if use_cache else None

        virtual_layer_idx = 0

        # Layer 0 (first layer, dense)
        if output_hidden_states:
            all_hidden_states += (hidden_states,)

        layer = self.layers[0]
        layer.self_attn.layer_idx = virtual_layer_idx

        layer_outputs = layer(
            hidden_states,
            attention_mask=attention_mask,
            position_ids=position_ids,
            past_key_value=past_key_values,
            output_attentions=output_attentions,
            use_cache=use_cache,
        )
        hidden_states = layer_outputs[0]

        if output_attentions:
            all_self_attns += (layer_outputs[1],)

        virtual_layer_idx += 1

        # Layer 1 (middle layer, MoE) - looped num_loop_iterations times
        for loop_iter in range(self.num_loop_iterations):
            if output_hidden_states:
                all_hidden_states += (hidden_states,)

            layer = self.layers[1]
            layer.self_attn.layer_idx = virtual_layer_idx

            layer_outputs = layer(
                hidden_states,
                attention_mask=attention_mask,
                position_ids=position_ids,
                past_key_value=past_key_values,
                output_attentions=output_attentions,
                use_cache=use_cache,
            )
            hidden_states = layer_outputs[0]

            if output_attentions:
                all_self_attns += (layer_outputs[1],)

            virtual_layer_idx += 1

        # Layer 2 (last layer, MoE)
        if output_hidden_states:
            all_hidden_states += (hidden_states,)

        layer = self.layers[2]
        layer.self_attn.layer_idx = virtual_layer_idx

        layer_outputs = layer(
            hidden_states,
            attention_mask=attention_mask,
            position_ids=position_ids,
            past_key_value=past_key_values,
            output_attentions=output_attentions,
            use_cache=use_cache,
        )
        hidden_states = layer_outputs[0]

        if output_attentions:
            all_self_attns += (layer_outputs[1],)

        hidden_states = self.norm(hidden_states)

        if output_hidden_states:
            all_hidden_states += (hidden_states,)

        if not return_dict:
            return tuple(
                v
                for v in [hidden_states, next_decoder_cache, all_hidden_states, all_self_attns]
                if v is not None
            )
        return BaseModelOutputWithPast(
            last_hidden_state=hidden_states,
            past_key_values=next_decoder_cache,
            hidden_states=all_hidden_states,
            attentions=all_self_attns,
        )


class ScatterbrainMoonlightForCausalLM(ScatterbrainMoonlightPreTrainedModel):
    _tied_weights_keys = ["lm_head.weight"]

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

        self.post_init()

    def get_input_embeddings(self):
        return self.model.embed_tokens

    def set_input_embeddings(self, value):
        self.model.embed_tokens = value

    def get_output_embeddings(self):
        return self.lm_head

    def set_output_embeddings(self, new_embeddings):
        self.lm_head = new_embeddings

    def set_decoder(self, decoder):
        self.model = decoder

    def get_decoder(self):
        return self.model

    def forward(
        self,
        input_ids: torch.LongTensor = None,
        attention_mask: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.LongTensor] = None,
        past_key_values: Optional[List[torch.FloatTensor]] = None,
        inputs_embeds: Optional[torch.FloatTensor] = None,
        labels: Optional[torch.LongTensor] = None,
        use_cache: Optional[bool] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
    ) -> Union[Tuple, CausalLMOutputWithPast]:
        output_attentions = (
            output_attentions
            if output_attentions is not None
            else self.config.output_attentions
        )
        output_hidden_states = (
            output_hidden_states
            if output_hidden_states is not None
            else self.config.output_hidden_states
        )
        return_dict = (
            return_dict if return_dict is not None else self.config.use_return_dict
        )

        outputs = 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,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )

        hidden_states = outputs[0]
        logits = self.lm_head(hidden_states)
        logits = logits.float()

        loss = None
        if labels is not None:
            shift_logits = logits[..., :-1, :].contiguous()
            shift_labels = labels[..., 1:].contiguous()
            loss_fct = CrossEntropyLoss()
            shift_logits = shift_logits.view(-1, self.config.vocab_size)
            shift_labels = shift_labels.view(-1)
            shift_labels = shift_labels.to(shift_logits.device)
            loss = loss_fct(shift_logits, shift_labels)

        if not return_dict:
            output = (logits,) + outputs[1:]
            return (loss,) + output if loss is not None else output

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

    def prepare_inputs_for_generation(
        self,
        input_ids,
        past_key_values=None,
        attention_mask=None,
        inputs_embeds=None,
        **kwargs,
    ):
        if past_key_values is not None:
            past_length = past_key_values.get_seq_length(0)

            if input_ids.shape[1] > past_length:
                remove_prefix_length = past_length
            else:
                remove_prefix_length = input_ids.shape[1] - 1

            input_ids = input_ids[:, remove_prefix_length:]

        position_ids = kwargs.get("position_ids", None)
        if attention_mask is not None and position_ids is None:
            position_ids = attention_mask.long().cumsum(-1) - 1
            position_ids.masked_fill_(attention_mask == 0, 1)
            if past_key_values is not None:
                position_ids = position_ids[:, -input_ids.shape[1] :]

        if inputs_embeds is not None and past_key_values is None:
            model_inputs = {"inputs_embeds": inputs_embeds}
        else:
            model_inputs = {"input_ids": input_ids}

        model_inputs.update(
            {
                "position_ids": position_ids,
                "past_key_values": past_key_values,
                "use_cache": kwargs.get("use_cache"),
                "attention_mask": attention_mask,
            }
        )
        return model_inputs

    @staticmethod
    def _reorder_cache(past_key_values, beam_idx):
        past_key_values.reorder_cache(beam_idx)
        return past_key_values