modeling_hybridna.py

# coding=utf-8
# Copyright 2024 AI21 Labs Ltd. and the HuggingFace Inc. team. All rights reserved.
#
# This code is based on EleutherAI's GPT-NeoX library and the GPT-NeoX
# and OPT implementations in this library. It has been modified from its
# original forms to accommodate minor architectural differences compared
# to GPT-NeoX and OPT used by the Meta AI team that trained the model.
#
# 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 HybriDNA model."""
import inspect
import math
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 einops import rearrange, repeat

from transformers.activations import ACT2FN
from transformers.cache_utils import DynamicCache  # we need __iter__ and __len__ of pkv
from transformers.modeling_attn_mask_utils import (
    AttentionMaskConverter,
)
from transformers.modeling_outputs import (
    BaseModelOutputWithPast,
    CausalLMOutputWithPast,
    SequenceClassifierOutputWithPast,
)
from transformers.modeling_utils import PreTrainedModel
from transformers.generation.utils import GenerationMixin
from transformers.utils import (
    add_start_docstrings,
    add_start_docstrings_to_model_forward,
    is_flash_attn_greater_or_equal_2_10,
    logging,
    replace_return_docstrings,
)
from transformers.utils.import_utils import (
    is_causal_conv1d_available,
    is_flash_attn_2_available,
    is_mamba_ssm_available,
)
from hf.configuration_hybridna import HybriDNAConfig
from mamba_ssm.ops.triton.layernorm_gated import RMSNorm as RMSNormGated

# try except block so it'll work with trust_remote_code.
try:
    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

    _flash_supports_window_size = "window_size" in list(inspect.signature(flash_attn_func).parameters)
except ImportError:
    pass


# try except block so it'll work with trust_remote_code.
try:
    from mamba_ssm.ops.selective_scan_interface import mamba_inner_fn, selective_scan_fn
    from mamba_ssm.ops.triton.selective_state_update import selective_state_update
    from mamba_ssm.ops.triton.ssd_combined import mamba_chunk_scan_combined, mamba_split_conv1d_scan_combined
except ImportError:
    selective_state_update, selective_scan_fn, mamba_inner_fn = None, None, None

# try except block so it'll work with trust_remote_code.
try:
    from causal_conv1d import causal_conv1d_fn, causal_conv1d_update
except ImportError:
    causal_conv1d_update, causal_conv1d_fn = None, None

is_fast_path_available = all(
    (selective_state_update, selective_scan_fn, causal_conv1d_fn, causal_conv1d_update, mamba_inner_fn)
)


logger = logging.get_logger(__name__)

_CONFIG_FOR_DOC = "HybriDNAConfig"


# Copied from transformers.models.llama.modeling_llama._get_unpad_data
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.int32), (1, 0))
    return (
        indices,
        cu_seqlens,
        max_seqlen_in_batch,
    )


# Copied from transformers.models.llama.modeling_llama.LlamaRMSNorm with Llama->HybriDNA
class HybriDNARMSNorm(nn.Module):
    def __init__(self, hidden_size, eps=1e-6):
        """
        HybriDNARMSNorm 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)


# Copied from transformers.models.llama.modeling_llama.repeat_kv
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)


class HybridMambaAttentionDynamicCache(DynamicCache):
    """
    A dynamic cache that can handle both the attention cache (which has a seq_len dimension) and the mamba cache
    (which has a constant shape regardless of seq_len).
    This cache has two sets of lists of tensors: `key_cache` and `value_cache` for attention cache and `conv_states`
    and `ssm_states` for mamba cache. Each of these lists has `num_layers` tensors. The expected shape for each tensor
    For attention layers, `key_cache` and `value_cache` have a shape of `(batch_size, num_heads, seq_len, head_dim)`,
    while `conv_states` and `ssm_states` have a shape of `(batch_size, 0)` (empty tensors).
    For mamba layers, `key_cache` and `value_cache` have a shape of `(batch_size, 0)` (empty tensors),
    while `conv_states` represents the convolution state and has a shape of `(batch_size, d_inner, d_conv)`,
    and `ssm_states` represents the ssm state and has a shape of `(batch_size, d_inner, d_state)`.
    """

    def __init__(self, config, batch_size, dtype=torch.float16, device=None):
        self.dtype = dtype
        self.layers_block_type = config.layers_block_type
        self.has_previous_state = False  # only used by mamba
        self.seq_offset = 0  # tracks current position in sequence for mamba2 cache
        intermediate_size = config.mamba_expand * config.hidden_size
        ssm_state_size = config.mamba_d_state
        conv_kernel_size = config.mamba_d_conv
        # Mamba2 uses different dimensions:
        # - conv_states includes x, B, C: intermediate_size + 2 * ssm_state_size
        # - ssm_states has shape (batch, num_heads, head_dim, ssm_state_size)
        num_heads = config.intermediate_size // config.head_dim
        head_dim = config.head_dim
        conv1d_dim = intermediate_size + 2 * ssm_state_size  # for xBC in Mamba2
        self.conv_states = []
        self.ssm_states = []
        for i in range(config.num_hidden_layers):
            if self.layers_block_type[i] == "mamba":
                self.conv_states += [
                    torch.zeros(batch_size, conv1d_dim, conv_kernel_size, device=device, dtype=dtype)
                ]
                self.ssm_states += [
                    torch.zeros(batch_size, num_heads, head_dim, ssm_state_size, device=device, dtype=dtype)
                ]
            else:
                self.conv_states += [torch.tensor([[]] * batch_size, device=device)]
                self.ssm_states += [torch.tensor([[]] * batch_size, device=device)]

        self.key_cache = [torch.tensor([[]] * batch_size, device=device) for _ in range(config.num_hidden_layers)]
        self.value_cache = [torch.tensor([[]] * batch_size, device=device) for _ in range(config.num_hidden_layers)]

    def update(
            self,
            key_states: torch.Tensor,
            value_states: torch.Tensor,
            layer_idx: int,
            cache_kwargs: Optional[Dict[str, Any]] = None,
    ) -> Tuple[torch.Tensor, torch.Tensor]:
        # Update the cache
        # Ensure tensors are contiguous for SDPA compatibility
        if self.key_cache[layer_idx].shape[-1] == 0:
            self.key_cache[layer_idx] = key_states.contiguous()
            self.value_cache[layer_idx] = value_states.contiguous()
        else:
            self.key_cache[layer_idx] = torch.cat([self.key_cache[layer_idx], key_states.contiguous()], dim=2)
            self.value_cache[layer_idx] = torch.cat([self.value_cache[layer_idx], value_states.contiguous()], dim=2)

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

    def reorder_cache(self, beam_idx: torch.LongTensor):
        """Reorders the cache for beam search, given the selected beam indices."""
        for layer_idx in range(len(self.key_cache)):
            device = self.key_cache[layer_idx].device
            self.key_cache[layer_idx] = self.key_cache[layer_idx].index_select(0, beam_idx.to(device))
            device = self.value_cache[layer_idx].device
            self.value_cache[layer_idx] = self.value_cache[layer_idx].index_select(0, beam_idx.to(device))

            device = self.conv_states[layer_idx].device
            self.conv_states[layer_idx] = self.conv_states[layer_idx].index_select(0, beam_idx.to(device))
            device = self.ssm_states[layer_idx].device
            self.ssm_states[layer_idx] = self.ssm_states[layer_idx].index_select(0, beam_idx.to(device))

    def __len__(self):
        """Return the number of layers in the cache."""
        return len(self.key_cache)

    def to_legacy_cache(self) -> Tuple[Tuple[torch.Tensor], Tuple[torch.Tensor]]:
        raise NotImplementedError("HybridMambaAttentionDynamicCache does not have a legacy cache equivalent.")

    @classmethod
    def from_legacy_cache(cls, past_key_values: Optional[Tuple[Tuple[torch.Tensor]]] = None) -> "DynamicCache":
        raise NotImplementedError("HybridMambaAttentionDynamicCache does not have a legacy cache equivalent.")


# Adapted from transformers.models.mistral.modeling_mistral.MistralAttention with Mistral->HybriDNA
class HybriDNAAttention(nn.Module):
    """
    Multi-headed attention from 'Attention Is All You Need' paper. Modified to use sliding window attention: Longformer
    and "Generating Long Sequences with Sparse Transformers".
    """

    def __init__(self, config: HybriDNAConfig, 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 a `layer_idx` is not recommended and will "
                "lead to errors during the forward call if caching is used. Please make sure to provide a `layer_idx` "
                "when creating this class."
            )

        self.hidden_size = config.hidden_size
        self.num_heads = config.num_attention_heads
        self.head_dim = self.hidden_size // self.num_heads
        self.num_key_value_heads = config.num_key_value_heads
        self.num_key_value_groups = self.num_heads // self.num_key_value_heads
        self.is_causal = True
        self.attention_dropout = config.attention_dropout

        if (self.head_dim * self.num_heads) != self.hidden_size:
            raise ValueError(
                f"hidden_size must be divisible by num_heads (got `hidden_size`: {self.hidden_size}"
                f" and `num_heads`: {self.num_heads})."
            )
        self.q_proj = nn.Linear(self.hidden_size, self.num_heads * self.head_dim, bias=False)
        self.k_proj = nn.Linear(self.hidden_size, self.num_key_value_heads * self.head_dim, bias=False)
        self.v_proj = nn.Linear(self.hidden_size, self.num_key_value_heads * self.head_dim, bias=False)
        self.o_proj = nn.Linear(self.num_heads * self.head_dim, self.hidden_size, bias=False)

    def forward(
            self,
            hidden_states: torch.Tensor,
            attention_mask: Optional[torch.Tensor] = None,
            position_ids: Optional[torch.LongTensor] = None,
            past_key_value: Optional[HybridMambaAttentionDynamicCache] = None,
            output_attentions: bool = False,
            use_cache: bool = False,
            cache_position: Optional[torch.LongTensor] = None,
    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
        bsz, q_len, _ = hidden_states.size()

        query_states = self.q_proj(hidden_states)
        key_states = self.k_proj(hidden_states)
        value_states = self.v_proj(hidden_states)

        query_states = query_states.view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
        key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
        value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)

        if past_key_value is not None:
            key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx)

        # repeat k/v heads if n_kv_heads < n_heads
        key_states = repeat_kv(key_states, self.num_key_value_groups)
        value_states = repeat_kv(value_states, self.num_key_value_groups)

        attn_weights = torch.matmul(query_states, key_states.transpose(2, 3)) / math.sqrt(self.head_dim)

        if attention_mask is not None:  # no matter the length, we just slice it
            causal_mask = attention_mask[:, :, :, : key_states.shape[-2]]
            attn_weights = attn_weights + causal_mask

        # upcast attention to fp32
        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.head_dim):
            raise ValueError(
                f"`attn_output` should be of size {(bsz, self.num_heads, q_len, self.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.hidden_size)

        attn_output = self.o_proj(attn_output)

        if not output_attentions:
            attn_weights = None

        return attn_output, attn_weights, past_key_value


# Adapted from transformers.models.mistral.modeling_mistral.MistralFlashAttention2 with Mistral->HybriDNA
class HybriDNAFlashAttention2(HybriDNAAttention):
    """
    HybriDNA flash attention module. This module inherits from `HybriDNAAttention` as the weights of the module stays
    untouched. The only required change would be on the forward pass where it needs to correctly call the public API of
    flash attention and deal with padding tokens in case the input contains any of them.
    """

    # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__
    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)

        # TODO: Should be removed once Flash Attention for RoCm is bumped to 2.1.
        # flash_attn<2.1 generates top-left aligned causal mask, while what is needed here is bottom-right alignment, that was made default for flash_attn>=2.1. This attribute is used to handle this difference. Reference: https://github.com/Dao-AILab/flash-attention/releases/tag/v2.1.0.
        # Beware that with flash_attn<2.1, using q_seqlen != k_seqlen (except for the case q_seqlen == 1) produces a wrong mask (top-left).
        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.Tensor] = None,
            position_ids: Optional[torch.LongTensor] = None,
            past_key_value: Optional[HybridMambaAttentionDynamicCache] = None,
            output_attentions: bool = False,
            use_cache: bool = False,
            cache_position: Optional[torch.LongTensor] = None,
            **kwargs,
    ):
        bsz, q_len, _ = hidden_states.size()

        query_states = self.q_proj(hidden_states)
        key_states = self.k_proj(hidden_states)
        value_states = self.v_proj(hidden_states)

        # Flash attention requires the input to have the shape
        # batch_size x seq_length x head_dim x hidden_dim
        # therefore we just need to keep the original shape
        query_states = query_states.view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
        key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
        value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)

        kv_seq_len = cache_position[-1]

        use_sliding_windows = (
                _flash_supports_window_size
                and getattr(self.config, "sliding_window", None) is not None
                and kv_seq_len > self.config.sliding_window
        )

        if not _flash_supports_window_size:
            logger.warning_once(
                "The current flash attention version does not support sliding window attention, for a more memory efficient implementation"
                " make sure to upgrade flash-attn library."
            )

        if past_key_value is not None:
            # Activate slicing cache only if the config has a value `sliding_windows` attribute
            cache_has_contents = cache_position[0] > 0
            if (
                    getattr(self.config, "sliding_window", None) is not None
                    and kv_seq_len > self.config.sliding_window
                    and cache_has_contents
            ):
                slicing_tokens = 1 - self.config.sliding_window

                past_key = past_key_value[self.layer_idx][0]
                past_value = past_key_value[self.layer_idx][1]

                past_key = past_key[:, :, slicing_tokens:, :].contiguous()
                past_value = past_value[:, :, slicing_tokens:, :].contiguous()

                if past_key.shape[-2] != self.config.sliding_window - 1:
                    raise ValueError(
                        f"past key must have a shape of (`batch_size, num_heads, self.config.sliding_window-1, head_dim`), got"
                        f" {past_key.shape}"
                    )

                if attention_mask is not None:
                    attention_mask = attention_mask[:, slicing_tokens:]
                    attention_mask = torch.cat([attention_mask, torch.ones_like(attention_mask[:, -1:])], dim=-1)

            key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx)

        # repeat k/v heads if n_kv_heads < n_heads
        key_states = repeat_kv(key_states, self.num_key_value_groups)
        value_states = repeat_kv(value_states, self.num_key_value_groups)
        dropout_rate = 0.0 if not self.training else self.attention_dropout

        # In PEFT, usually we cast the layer norms in float32 for training stability reasons
        # therefore the input hidden states gets silently casted in float32. Hence, we need
        # cast them back in float16 just to be sure everything works as expected.
        input_dtype = query_states.dtype
        if input_dtype == torch.float32:
            if torch.is_autocast_enabled():
                target_dtype = torch.get_autocast_gpu_dtype()
            # Handle the case where the model is quantized
            elif hasattr(self.config, "_pre_quantization_dtype"):
                target_dtype = self.config._pre_quantization_dtype
            else:
                target_dtype = self.q_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)

        # Reshape to the expected shape for Flash Attention
        query_states = query_states.transpose(1, 2)
        key_states = key_states.transpose(1, 2)
        value_states = value_states.transpose(1, 2)

        attn_output = self._flash_attention_forward(
            query_states,
            key_states,
            value_states,
            attention_mask,
            q_len,
            dropout=dropout_rate,
            use_sliding_windows=use_sliding_windows,
        )

        attn_output = attn_output.reshape(bsz, q_len, self.hidden_size).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,
            use_sliding_windows=False,
    ):
        """
        Calls the forward method of Flash Attention - if the input hidden states contain at least one padding token
        first unpad the input, then computes the attention scores and pad the final attention scores.
        Args:
            query_states (`torch.Tensor`):
                Input query states to be passed to Flash Attention API
            key_states (`torch.Tensor`):
                Input key states to be passed to Flash Attention API
            value_states (`torch.Tensor`):
                Input value states to be passed to Flash Attention API
            attention_mask (`torch.Tensor`):
                The padding mask - corresponds to a tensor of size `(batch_size, seq_len)` where 0 stands for the
                position of padding tokens and 1 for the position of non-padding tokens.
            dropout (`float`, *optional*):
                Attention dropout
            softmax_scale (`float`, *optional*):
                The scaling of QK^T before applying softmax. Default to 1 / sqrt(head_dim)
            use_sliding_windows (`bool`, *optional*):
                Whether to activate sliding window attention.
        """
        if not self._flash_attn_uses_top_left_mask:
            causal = self.is_causal
        else:
            # TODO: Remove the `query_length != 1` check once Flash Attention for RoCm is bumped to 2.1. For details, please see the comment in LlamaFlashAttention2 __init__.
            causal = self.is_causal and query_length != 1

        # Contains at least one padding token in the sequence
        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

            if not use_sliding_windows:
                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,
                )
            else:
                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,
                    window_size=(self.config.sliding_window, self.config.sliding_window),
                )

            attn_output = pad_input(attn_output_unpad, indices_q, batch_size, query_length)
        else:
            if not use_sliding_windows:
                attn_output = flash_attn_func(
                    query_states,
                    key_states,
                    value_states,
                    dropout,
                    softmax_scale=softmax_scale,
                    causal=causal,
                )
            else:
                attn_output = flash_attn_func(
                    query_states,
                    key_states,
                    value_states,
                    dropout,
                    softmax_scale=softmax_scale,
                    causal=causal,
                    window_size=(self.config.sliding_window, self.config.sliding_window),
                )

        return attn_output

    # Copied from transformers.models.mixtral.modeling_mixtral.MixtralFlashAttention2._upad_input
    def _upad_input(self, query_layer, key_layer, value_layer, attention_mask, query_length):
        batch_size, kv_seq_len, num_heads, head_dim = key_layer.shape

        # On the first iteration we need to properly re-create the padding mask
        # by slicing it on the proper place
        if kv_seq_len != attention_mask.shape[-1]:
            attention_mask_num_tokens = attention_mask.shape[-1]
            attention_mask = attention_mask[:, attention_mask_num_tokens - kv_seq_len :]

        indices_k, cu_seqlens_k, max_seqlen_in_batch_k = _get_unpad_data(attention_mask)

        key_layer = index_first_axis(key_layer.reshape(batch_size * kv_seq_len, num_heads, head_dim), indices_k)
        value_layer = index_first_axis(value_layer.reshape(batch_size * kv_seq_len, num_heads, head_dim), indices_k)

        if query_length == kv_seq_len:
            query_layer = index_first_axis(
                query_layer.reshape(batch_size * kv_seq_len, 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
            )  # There is a memcpy here, that is very bad.
            indices_q = cu_seqlens_q[:-1]
            query_layer = query_layer.squeeze(1)
        else:
            # The -q_len: slice assumes left padding.
            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),
        )


# Adapted from transformers.models.mistral.modeling_mistral.MistralSdpaAttention with Mistral->HybriDNA
class HybriDNASdpaAttention(HybriDNAAttention):
    """
    HybriDNA attention module using torch.nn.functional.scaled_dot_product_attention. This module inherits from
    `HybriDNAAttention` as the weights of the module stays untouched. The only changes are on the forward pass to adapt to
    SDPA API.
    """

    # Adapted from HybriDNAAttention.forward
    def forward(
            self,
            hidden_states: torch.Tensor,
            attention_mask: Optional[torch.Tensor] = None,
            position_ids: Optional[torch.LongTensor] = None,
            past_key_value: Optional[HybridMambaAttentionDynamicCache] = None,
            output_attentions: bool = False,
            use_cache: bool = False,
            cache_position: Optional[torch.LongTensor] = None,
    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
        if output_attentions:
            # TODO: Improve this warning with e.g. `model.config.attn_implementation = "manual"` once this is implemented.
            logger.warning_once(
                "HybriDNAModel is using HybriDNASdpaAttention, but `torch.nn.functional.scaled_dot_product_attention` does not support `output_attentions=True`. Falling back to the manual attention implementation, "
                'but specifying the manual implementation will be required from Transformers version v5.0.0 onwards. This warning can be removed using the argument `attn_implementation="eager"` when loading the model.'
            )
            return super().forward(
                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,
            )

        bsz, q_len, _ = hidden_states.size()

        query_states = self.q_proj(hidden_states)
        key_states = self.k_proj(hidden_states)
        value_states = self.v_proj(hidden_states)

        query_states = query_states.view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
        key_states = key_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
        value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)

        if past_key_value is not None:
            key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx)

        key_states = repeat_kv(key_states, self.num_key_value_groups)
        value_states = repeat_kv(value_states, self.num_key_value_groups)

        causal_mask = attention_mask
        if attention_mask is not None:
            causal_mask = causal_mask[:, :, :, : key_states.shape[-2]]

        # SDPA with memory-efficient backend is currently (torch==2.1.2) bugged with non-contiguous inputs with custom attn_mask,
        # Reference: https://github.com/pytorch/pytorch/issues/112577.
        if query_states.device.type == "cuda" and attention_mask is not None:
            query_states = query_states.contiguous()
            key_states = key_states.contiguous()
            value_states = value_states.contiguous()
            causal_mask = causal_mask.contiguous()

        attn_output = torch.nn.functional.scaled_dot_product_attention(
            query_states,
            key_states,
            value_states,
            attn_mask=causal_mask,
            dropout_p=self.attention_dropout if self.training else 0.0,
            # The q_len > 1 is necessary to match with AttentionMaskConverter.to_causal_4d that does not create a causal mask in case q_len == 1.
            is_causal=self.is_causal and attention_mask is None and q_len > 1,
        )

        attn_output = attn_output.transpose(1, 2).contiguous()
        attn_output = attn_output.view(bsz, q_len, self.hidden_size)

        attn_output = self.o_proj(attn_output)

        return attn_output, None, past_key_value


HYBRIDNA_ATTENTION_CLASSES = {
    "eager": HybriDNAAttention,
    "flash_attention_2": HybriDNAFlashAttention2,
    "sdpa": HybriDNASdpaAttention,
}

class HybriDNAMamba2RMSNorm(nn.Module):
    def __init__(self, hidden_size, eps=1e-6, normalize=False):
        """
        HybriDNAMamba2RMSNorm is equivalent to T5LayerNorm and LlamaRMSNorm but with optional residual normalizing
        """
        super().__init__()
        self.weight = nn.Parameter(torch.ones(hidden_size))
        self.eps = eps
        self.normalize = normalize

    def forward(self, hidden_states, residual=None):
        input_dtype = hidden_states.dtype
        hidden_states = hidden_states.to(torch.float32)

        # residual normalization introduced for instability, see section 7 of the paper
        if residual is not None and self.normalize:
            hidden_states = hidden_states * nn.functional.silu(residual.to(torch.float32))

        variance = hidden_states.pow(2).mean(-1, keepdim=True)
        hidden_states = hidden_states * torch.rsqrt(variance + self.eps)
        hidden_states = hidden_states * self.weight

        return hidden_states.to(input_dtype)

class HybriDNAMamba2Mixer(nn.Module):
    """
    Using the found relation to the attention mechanism under certain conditions (State-Space-Duality SSD),
    we use the Multi-input SSM which can be seen as a counterpart to the Multi-value Attention with analogues:
    - X ~= V
    - B ~= Q
    - C ~= K
    - A (1-SS(a)) ~= Attention Mask

    For an overview, see the mamba2 paper, section 6, figure 4.
    """

    def __init__(self, config: HybriDNAConfig, layer_idx: int):
        super().__init__()
        self.hidden_size = config.hidden_size
        self.ssm_state_size = config.mamba_d_state
        self.conv_kernel_size = config.mamba_d_conv
        self.intermediate_size = config.mamba_expand * config.hidden_size
        self.head_dim = config.head_dim
        self.num_heads = config.intermediate_size // self.head_dim
        self.chunk_size = config.chunk_size
        self.dt_min = 0
        self.dt_max = float("inf")
        self.layer_idx = layer_idx
        self.use_bias = config.mamba_proj_bias
        self.use_conv_bias = config.mamba_conv_bias
        self.use_triton_kernels = config.use_mamba_kernels

        # Parallel projection of the input hidden states
        self.in_proj = nn.Linear(
            in_features=self.hidden_size,
            out_features=2 * (self.intermediate_size + self.ssm_state_size) + self.num_heads,
            bias=self.use_bias,
        )

        conv1d_dim = self.intermediate_size + 2 * self.ssm_state_size
        self.conv1d = nn.Conv1d(
            in_channels=conv1d_dim,
            out_channels=conv1d_dim,
            bias=self.use_conv_bias,
            kernel_size=config.mamba_d_conv,
            groups=conv1d_dim,
            padding=config.mamba_d_conv - 1,
        )

        self.activation = config.hidden_act
        self.act = ACT2FN[config.hidden_act]

        # We only use a bias as parameter
        self.dt_bias = nn.Parameter(torch.rand(size=(self.num_heads,)))

        # Scalar initialization of A, i.e. 1-Semi-Separable Matrix of A (== 1-SS(a))
        A = torch.empty(self.num_heads, dtype=torch.float32).uniform_(1,16)
        self.A_log = nn.Parameter(torch.log(A))

        # As D is a skip connection with A, it is also a scalar of the same shape as A
        self.D = nn.Parameter(torch.ones(self.num_heads))

        # Residual normalization introduced for instability, see section 7 of the paper
        self.norm = HybriDNAMamba2RMSNorm(self.intermediate_size, eps=1e-5, normalize=True)

        self.out_proj = nn.Linear(self.intermediate_size, self.hidden_size, bias=self.use_bias)

    def _conv1d(self, xBC, seq_len, use_triton_kernels, cache, cached_start, cached_forward):
        # Init cache with first "real" values
        if cached_start:
            xBC_t = rearrange(xBC, "b l d -> b d l")
            cache.conv_states[self.layer_idx].copy_(
                nn.functional.pad(xBC_t, (self.conv_kernel_size - xBC_t.shape[-1], 0))
            )

        if is_fast_path_available and use_triton_kernels:
            if cached_forward:
                # causal_conv1d_update expects (batch, dim) for single token
                xBC = causal_conv1d_update(
                    xBC.squeeze(1),  # (batch, 1, dim) -> (batch, dim)
                    cache.conv_states[self.layer_idx],
                    rearrange(self.conv1d.weight, "d 1 w -> d w"),
                    self.conv1d.bias,
                    self.activation,
                )
                xBC = xBC.unsqueeze(1)  # (batch, dim) -> (batch, 1, dim)
            else:
                xBC = causal_conv1d_fn(
                    xBC.transpose(1, 2),
                    rearrange(self.conv1d.weight, "d 1 w -> d w"),
                    bias=self.conv1d.bias,
                    activation=self.activation,
                ).transpose(1, 2)
        else:
            if cached_forward:
                cache.conv_states[self.layer_idx].copy_(
                    torch.roll(cache.conv_states[self.layer_idx], shifts=-1, dims=-1)
                )
                cache.conv_states[self.layer_idx][:, :, -1] = xBC
                xBC = torch.sum(
                    cache.conv_states[self.layer_idx] * rearrange(self.conv1d.weight, "d 1 w -> d w"), dim=-1
                )
                if self.conv1d.bias is not None:
                    xBC = xBC + self.conv1d.bias
                xBC = self.act(xBC)
            else:
                xBC = self.act(self.conv1d(xBC.transpose(1, 2))[..., :seq_len].transpose(1, 2))

        return xBC

    def _ssd_naive(self, x, dt, A, B, C, chunk_size, dt_min, dt_max, initial_states=None, return_final_states=False):
        """
        Arguments:
            x:  (batch_size, seq_len, num_heads, head_dim)
            dt: (batch_size, seq_len, num_heads)
            A:  (num_heads)
            B:  (batch_size, seq_len, num_heads, ssm_state_size)
            C:  (batch_size, seq_len, num_heads, ssm_state_size)
        Return:
            y:  (batch_size, seq_len, num_heads, head_dim)
        """

        def pad_by_size(x, pad_size):
            """
            Padding x tensor with `pad_size` on the seq_len dim (dim=1)

            Assumes that we only have tensors of either size 4 or 3
            """
            assert 2 < len(x.shape) < 5

            pad_shape = (0, 0, 0, 0, 0, pad_size, 0, 0) if len(x.shape) == 4 else (0, 0, 0, pad_size, 0, 0)

            return nn.functional.pad(x, pad_shape, mode="constant", value=0)

        def segsum(x):
            """
            More stable segment sum calculation
            """
            T = x.size(-1)
            x = repeat(x, "... d -> ... d e", e=T)
            mask = torch.tril(torch.ones(T, T, device=x.device, dtype=torch.bool), diagonal=-1)
            x = x.masked_fill(~mask, 0)
            x_segsum = torch.cumsum(x, dim=-2)
            mask = torch.tril(torch.ones(T, T, device=x.device, dtype=torch.bool), diagonal=0)
            x_segsum = x_segsum.masked_fill(~mask, -torch.inf)
            return x_segsum

        # Since it is parallelized by chunks they have to be of the same size which we ensure by padding
        seq_len = x.shape[1]
        pad_size = chunk_size - (seq_len % chunk_size)

        # dt softplus and clamping
        dt = nn.functional.softplus(dt + self.dt_bias)
        dt = torch.clamp(dt, dt_min, dt_max)

        D_residual = self.D.unsqueeze(-1) * pad_by_size(x, pad_size)

        # Discretize x and A
        x = x * dt.unsqueeze(-1)
        A = A.to(x.dtype) * dt

        # Rearrange into blocks/chunks
        x, A, B, C = [
            rearrange(pad_by_size(t, pad_size), "b (c l) ... -> b c l ...", l=chunk_size) for t in (x, A, B, C)
        ]

        A = rearrange(A, "b c l h -> b h c l")
        A_cumsum = torch.cumsum(A, dim=-1)

        # 1. Compute the output for each intra-chunk (diagonal blocks)
        L = torch.exp(segsum(A))
        Y_diag = torch.einsum("bclhn,bcshn,bhcls,bcshp->bclhp", C, B, L, x)

        # 2. Compute the state for each intra-chunk
        # (right term of low-rank factorization of off-diagonal blocks; B terms)
        decay_states = torch.exp((A_cumsum[:, :, :, -1:] - A_cumsum))
        states = torch.einsum("bclhn,bhcl,bclhp->bchpn", B, decay_states, x)

        # 3. Compute the inter-chunk SSM recurrence; produces correct SSM states at chunk boundaries
        # (middle term of factorization of off-diag blocks; A terms)
        if initial_states is None:
            initial_states = torch.zeros_like(states[:, :1])
        states = torch.cat([initial_states, states], dim=1)
        decay_chunk = torch.exp(segsum(nn.functional.pad(A_cumsum[:, :, :, -1], (1, 0))))
        new_states = torch.einsum("bhzc,bchpn->bzhpn", decay_chunk, states)
        states, final_state = new_states[:, :-1], new_states[:, -1]

        # 4. Compute state -> output conversion per chunk
        # (left term of low-rank factorization of off-diagonal blocks; C terms)
        state_decay_out = torch.exp(A_cumsum)
        Y_off = torch.einsum("bclhn,bchpn,bhcl->bclhp", C, states, state_decay_out)

        # Add output of intra-chunk and inter-chunk terms (diagonal and off-diagonal blocks)
        y = rearrange(Y_diag + Y_off, "b c l h p -> b (c l) h p")

        # Add D residual to final output
        y = y + D_residual

        # Cutting off padded chunks
        if pad_size > 0:
            y = y[:, :seq_len, :, :]

        if not return_final_states:
            return y
        else:
            return y, final_state

    def _ssd(
            self, x, B, C, dt, initial_state, return_final_state, use_triton_kernels, cache, cached_start, cached_forward
    ):
        # Discretize 1-SS(a)
        A = -torch.exp(self.A_log.float())  # .float() to avoid infs/nans

        last_state = None
        if not cached_forward:
            if use_triton_kernels:
                y = mamba_chunk_scan_combined(
                    x=rearrange(x, pattern="b l (h p) -> b l h p", p=self.head_dim),
                    dt=dt,
                    A=A,
                    B=rearrange(B, pattern="b l n -> b l 1 n"),
                    C=rearrange(C, pattern="b l n -> b l 1 n"),
                    chunk_size=self.chunk_size,
                    D=self.D,
                    z=None,
                    initial_states=initial_state,
                    dt_bias=self.dt_bias,
                    dt_softplus=True,
                    seq_idx=None,
                    dt_limit=(self.dt_min, self.dt_max),
                    return_final_states=cached_start or return_final_state,
                )
            else:
                initial_state = rearrange(initial_state, "b n h p -> b 1 n h p") if initial_state is not None else None
                y = self._ssd_naive(
                    x=rearrange(x, pattern="b l (h p) -> b l h p", p=self.head_dim),
                    dt=dt,
                    A=A,
                    B=rearrange(B, pattern="b l n -> b l 1 n"),
                    C=rearrange(C, pattern="b l n -> b l 1 n"),
                    chunk_size=self.chunk_size,
                    initial_states=initial_state,
                    dt_min=self.dt_min,
                    dt_max=self.dt_max,
                    return_final_states=cached_start or return_final_state,
                )
            if cached_start or return_final_state:
                y, last_state = y
                if cached_start:
                    cache.ssm_states[self.layer_idx].copy_(last_state)

            y = rearrange(y, "b l h p -> b l (h p)")
        else:
            # For cached_forward, squeeze the sequence dimension (seq_len=1)
            x = x.squeeze(1)  # (batch, 1, dim) -> (batch, dim)
            B = B.squeeze(1)  # (batch, 1, ssm_state_size) -> (batch, ssm_state_size)
            C = C.squeeze(1)  # (batch, 1, ssm_state_size) -> (batch, ssm_state_size)
            dt = dt.squeeze(1)  # (batch, 1, num_heads) -> (batch, num_heads)

            if use_triton_kernels:
                # Preparing values for single step
                A = repeat(A, "h -> h p n", p=self.head_dim, n=self.ssm_state_size).to(dtype=torch.float32)
                dt = repeat(dt, "b h -> b h p", p=self.head_dim)
                dt_bias = repeat(self.dt_bias, "h -> h p", p=self.head_dim)
                D = repeat(self.D, "h -> h p", p=self.head_dim)
                x_reshaped = rearrange(x, "b (h p) -> b h p", p=self.head_dim)

                # Triton kernel for updating states in-place and returning the hidden state
                y = selective_state_update(
                    state=cache.ssm_states[self.layer_idx],
                    x=x_reshaped,
                    dt=dt,
                    A=A,
                    B=B,
                    C=C,
                    D=D,
                    z=None,
                    dt_bias=dt_bias,
                    dt_softplus=True,
                )
            else:
                # Get time step with softplus and bias
                dt = nn.functional.softplus(dt + self.dt_bias.to(dtype=dt.dtype))
                # dt is already squeezed to (batch, num_heads)

                # Discretize A
                dA = torch.exp(dt * A)

                # Discretize B and x
                x = rearrange(x, "b (h p) -> b h p", p=self.head_dim)
                dBx = torch.einsum("bh,bn,bhp->bhpn", dt, B, x)

                # State calculation
                cache.ssm_states[self.layer_idx].copy_(
                    cache.ssm_states[self.layer_idx] * rearrange(dA, "b h -> b h 1 1") + dBx
                )

                # Subsequent output
                y = torch.einsum("bhpn,bn->bhp", cache.ssm_states[self.layer_idx], C)

                # D skip connection
                y = y + rearrange(self.D, "h -> h 1") * x

            # Reshaping to have seq_len == 1
            y = rearrange(y, "b h p -> b 1 (h p)")

            # Optional output of last state
            if return_final_state:
                last_state = cache.ssm_states[self.layer_idx].clone()

        return y, last_state

    def _forward(
            self,
            hidden_states,
            use_triton_kernels,
            initial_state=None,
            return_final_state=False,
            cache: Optional[HybridMambaAttentionDynamicCache] = None,
    ):
        # Managing cache state
        if cache is not None:
            cached_start = cache.seq_offset == 0
            cached_forward = not cached_start
        else:
            cached_start = False
            cached_forward = False

        # Supporting cached values as well as passing initial states but not both at the same time
        if initial_state is not None and cached_forward:
            raise ValueError("Subsequent caching and passing initial states is not possible at the same time!")

        # 1. Parallel projection for the input
        zxbcdt = self.in_proj(hidden_states)

        # 2-5. Training combined into one triton kernel
        if self.training and cache is None and is_fast_path_available and use_triton_kernels:
            y = mamba_split_conv1d_scan_combined(
                zxbcdt=zxbcdt,
                conv1d_weight=rearrange(self.conv1d.weight, "d 1 w -> d w"),
                conv1d_bias=self.conv1d.bias,
                dt_bias=self.dt_bias,
                A=-torch.exp(self.A_log),
                D=self.D,
                chunk_size=self.chunk_size,
                seq_idx=None,
                activation=self.activation,
                rmsnorm_weight=self.norm.weight,
                rmsnorm_eps=self.norm.eps,
                outproj_weight=self.out_proj.weight,
                outproj_bias=self.out_proj.bias,
                headdim=self.head_dim,
                ngroups=1,
                norm_before_gate=False,  # not the same as our variant's normalization var
                dt_limit=(self.dt_min, self.dt_max),
                initial_states=initial_state,
                return_final_states=return_final_state,
            )
            last_state = None
            if return_final_state:
                y, last_state = y
            return y

        # Reconstructing the necessary vars
        d_mlp = (zxbcdt.shape[-1] - 2 * self.intermediate_size - 2 * self.ssm_state_size - self.num_heads) // 2
        z0, x0, z, xBC, dt = torch.split(
            zxbcdt,
            [d_mlp, d_mlp, self.intermediate_size, self.intermediate_size + 2 * self.ssm_state_size, self.num_heads],
            dim=-1,
        )

        # 2. Causal convolution for partial set of variables ("input", B, C)
        xBC = self._conv1d(
            xBC=xBC,
            seq_len=hidden_states.shape[1],
            use_triton_kernels=use_triton_kernels,
            cache=cache,
            cached_start=cached_start,
            cached_forward=cached_forward,
        )

        # Reconstruct causal convolution vars
        x, B, C = torch.split(xBC, [self.intermediate_size, self.ssm_state_size, self.ssm_state_size], dim=-1)

        # 3. State Space Duality (SSD)
        y, last_state = self._ssd(
            x=x,
            B=B,
            C=C,
            dt=dt,
            initial_state=initial_state,
            return_final_state=return_final_state,
            use_triton_kernels=use_triton_kernels,
            cache=cache,
            cached_start=cached_start,
            cached_forward=cached_forward,
        )

        # 4. Gate normalization introduced for instability, see section 7 of the paper
        y = self.norm(y, residual=z)
        if d_mlp > 0:
            y = torch.cat([self.act(z0) * x0, y], dim=-1)

        # 5. Out projecting
        y = self.out_proj(y)

        return y

    def forward(
            self, hidden_states, initial_state=None, return_final_state=False, cache_params: HybridMambaAttentionDynamicCache = None
    ):
        use_triton_kernels = "cuda" in self.in_proj.weight.device.type and self.use_triton_kernels

        # AMD might be available later on with https://github.com/state-spaces/mamba/pull/359
        if use_triton_kernels:
            if not is_fast_path_available:
                logger.warning_once(
                    "Faster path is not available because `(causal_conv1d_fn, causal_conv1d_update)` is None. "
                    "Falling back to slower implementation. To install follow https://github.com/Dao-AILab/causal-conv1d"
                )
        else:
            logger.warning_once(
                "Fast path is not available because the GPU is not properly utilized. "
                "Falling back to naive implementation."
            )
        return self._forward(hidden_states, use_triton_kernels, initial_state, return_final_state, cache_params)


# Copied from transformers.models.mistral.modeling_mistral.MistralMLP with Mistral->HybriDNA
class HybriDNAMLP(nn.Module):
    def __init__(self, config):
        super().__init__()
        self.config = config
        self.hidden_size = config.hidden_size
        self.intermediate_size = config.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):
        return self.down_proj(self.act_fn(self.gate_proj(x)) * self.up_proj(x))


class HybriDNAAttentionDecoderLayer(nn.Module):
    def __init__(self, config: HybriDNAConfig, layer_idx: int):
        super().__init__()
        self.self_attn = HYBRIDNA_ATTENTION_CLASSES[config._attn_implementation](config, layer_idx)
        self.feed_forward = HybriDNAMLP(config)
        self.input_layernorm = HybriDNARMSNorm(config.hidden_size, eps=config.rms_norm_eps)
        self.pre_ff_layernorm = HybriDNARMSNorm(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[HybridMambaAttentionDynamicCache] = None,
                output_attentions: Optional[bool] = False,
                use_cache: Optional[bool] = False, cache_position: Optional[torch.LongTensor] = None
                ) -> Tuple[torch.Tensor, Optional[Tuple[torch.Tensor, torch.Tensor]]]:
        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,
            cache_position=cache_position,
        )

        # residual connection after attention
        hidden_states = residual + hidden_states

        # feed-forward
        residual = hidden_states
        hidden_states = self.pre_ff_layernorm(hidden_states)
        hidden_states = self.feed_forward(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


class HybriDNAMambaDecoderLayer(nn.Module):
    def __init__(self, config: HybriDNAConfig, layer_idx: int):
        super().__init__()
        self.mamba = HybriDNAMamba2Mixer(config=config, layer_idx=layer_idx)
        self.feed_forward = HybriDNAMLP(config)
        self.input_layernorm = HybriDNARMSNorm(config.hidden_size, eps=config.rms_norm_eps)
        self.pre_ff_layernorm = HybriDNARMSNorm(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[HybridMambaAttentionDynamicCache] = None,
                output_attentions: Optional[bool] = False,
                use_cache: Optional[bool] = False, cache_position: Optional[torch.LongTensor] = None
                ) -> Tuple[torch.Tensor, Optional[Tuple[torch.Tensor, torch.Tensor]]]:
        residual = hidden_states

        hidden_states = self.input_layernorm(hidden_states)

        hidden_states = self.mamba(
            hidden_states=hidden_states,
            cache_params=past_key_value,
        )
        self_attn_weights = None

        # residual connection after mamba
        hidden_states = residual + hidden_states

        # feed-forward
        residual = hidden_states
        hidden_states = self.pre_ff_layernorm(hidden_states)
        hidden_states = self.feed_forward(hidden_states)
        hidden_states = residual + hidden_states

        outputs = (hidden_states,)
        if output_attentions:
            outputs += (self_attn_weights,)
        if use_cache:
            outputs += (past_key_value,)
        return outputs


HYBRIDNA_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.)
    This model is also a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) subclass.
    Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage
    and behavior.
    Parameters:
        config ([`HybriDNAConfig`]):
            Model configuration class with all the parameters of the model. Initializing with a config file does not
            load the weights associated with the model, only the configuration. Check out the
            [`~PreTrainedModel.from_pretrained`] method to load the model weights.
"""


@add_start_docstrings(
    "The bare HybriDNA Model outputting raw hidden-states without any specific head on top.",
    HYBRIDNA_START_DOCSTRING,
)
class HybriDNAPreTrainedModel(PreTrainedModel):
    config_class = HybriDNAConfig
    base_model_prefix = "model"
    supports_gradient_checkpointing = True
    _no_split_modules = ["HybriDNAAttentionDecoderLayer", "HybriDNAMambaDecoderLayer"]
    _skip_keys_device_placement = "past_key_values"
    _supports_flash_attn_2 = True
    _supports_sdpa = True
    _supports_cache_class = True

    def _init_weights(self, module):
        std = self.config.initializer_range
        if isinstance(module, (nn.Linear, nn.Conv1d)):
            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_()


HYBRIDNA_INPUTS_DOCSTRING = r"""
    Args:
        input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
            Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you provide
            it.
            Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and
            [`PreTrainedTokenizer.__call__`] for details.
            [What are input IDs?](../glossary#input-ids)
        attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*):
            Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
            - 1 for tokens that are **not masked**,
            - 0 for tokens that are **masked**.
            [What are attention masks?](../glossary#attention-mask)
            Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and
            [`PreTrainedTokenizer.__call__`] for details.
            If `past_key_values` is used, optionally only the last `input_ids` have to be input (see
            `past_key_values`).
            If you want to change padding behavior, you should read [`modeling_opt._prepare_decoder_attention_mask`]
            and modify to your needs. See diagram 1 in [the paper](https://arxiv.org/abs/1910.13461) for more
            information on the default strategy.
            - 1 indicates the head is **not masked**,
            - 0 indicates the head is **masked**.
        position_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
            Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0,
            config.n_positions - 1]`.
            [What are position IDs?](../glossary#position-ids)
        past_key_values (`HybridMambaAttentionDynamicCache`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
            A HybridMambaAttentionDynamicCache object containing pre-computed hidden-states (keys and values in the
            self-attention blocks and convolution and ssm states in the mamba blocks) that can be used (see
            `past_key_values` input) to speed up sequential decoding.
            Key and value cache tensors have shape `(batch_size, num_heads, seq_len, head_dim)`.
            Convolution and ssm states tensors have shape `(batch_size, d_inner, d_conv)` and
            `(batch_size, d_inner, d_state)` respectively.
            See the `HybridMambaAttentionDynamicCache` class for more details.
            If `past_key_values` are used, the user can optionally input only the last `input_ids` (those that
            don't have their past key value states given to this model) of shape `(batch_size, 1)` instead of all
            `input_ids` of shape `(batch_size, sequence_length)`.
        inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
            Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. This
            is useful if you want more control over how to convert `input_ids` indices into associated vectors than the
            model's internal embedding lookup matrix.
        use_cache (`bool`, *optional*):
            If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see
            `past_key_values`).
        output_attentions (`bool`, *optional*):
            Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned
            tensors for more detail.
        output_hidden_states (`bool`, *optional*):
            Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
            more detail.
        return_dict (`bool`, *optional*):
            Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
        cache_position (`torch.LongTensor` of shape `(sequence_length)`, *optional*):
            Indices depicting the position of the input sequence tokens in the sequence. Contrarily to `position_ids`,
            this tensor is not affected by padding. It is used to update the cache in the correct position and to infer
            the complete sequence length.
"""

ALL_DECODER_LAYER_TYPES = {"attention": HybriDNAAttentionDecoderLayer, "mamba": HybriDNAMambaDecoderLayer}


@add_start_docstrings(
    "The bare HybriDNA Model outputting raw hidden-states without any specific head on top.",
    HYBRIDNA_START_DOCSTRING,
)
class HybriDNAModel(HybriDNAPreTrainedModel):
    """
    Transformer decoder consisting of *config.num_hidden_layers* layers. Each layer is a [`HybriDNADecoderLayer`]
    Args:
        config: HybriDNAConfig
    """

    def __init__(self, config: HybriDNAConfig):
        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)
        decoder_layers = []
        for i in range(config.num_hidden_layers):
            layer_class = ALL_DECODER_LAYER_TYPES[config.layers_block_type[i]]
            decoder_layers.append(layer_class(config, layer_idx=i))
        self.layers = nn.ModuleList(decoder_layers)

        self._attn_implementation = config._attn_implementation
        self.final_layernorm = HybriDNARMSNorm(config.hidden_size, eps=config.rms_norm_eps)

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

    def get_input_embeddings(self):
        return self.embed_tokens

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

    @add_start_docstrings_to_model_forward(HYBRIDNA_INPUTS_DOCSTRING)
    def forward(
            self,
            input_ids: torch.LongTensor = None,
            attention_mask: Optional[torch.Tensor] = None,
            position_ids: Optional[torch.LongTensor] = None,
            past_key_values: Optional[HybridMambaAttentionDynamicCache] = None,
            inputs_embeds: Optional[torch.Tensor] = None,
            use_cache: Optional[bool] = None,
            output_attentions: Optional[bool] = None,
            output_hidden_states: Optional[bool] = None,
            return_dict: Optional[bool] = None,
            cache_position: Optional[torch.LongTensor] = 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 None) ^ (inputs_embeds is not None):
            raise ValueError(
                "You cannot specify both input_ids and inputs_embeds at the same time, and must specify either one"
            )

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

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

        if use_cache and past_key_values is None:
            logger.warning_once(
                "HybriDNA requires an initialized `HybridMambaAttentionDynamicCache` to return a cache. None was "
                "provided, so no cache will be returned."
            )

        if cache_position is None:
            past_seen_tokens = past_key_values.seq_offset if past_key_values is not None else 0
            cache_position = torch.arange(past_seen_tokens, past_seen_tokens + hidden_states.shape[1], device=hidden_states.device)
        if position_ids is None:
            position_ids = cache_position.unsqueeze(0)

        causal_mask = self._update_causal_mask(attention_mask, inputs_embeds, cache_position)

        all_hidden_states = () if output_hidden_states else None
        all_self_attns = () if output_attentions else None

        for decoder_layer in self.layers:
            if output_hidden_states:
                all_hidden_states += (hidden_states,)

            if self.gradient_checkpointing and self.training:
                layer_outputs = self._gradient_checkpointing_func(
                    decoder_layer.__call__,
                    hidden_states,
                    causal_mask,
                    position_ids,
                    past_key_values,
                    output_attentions,
                    use_cache,
                    cache_position,
                )
            else:
                layer_outputs = decoder_layer(
                    hidden_states,
                    attention_mask=causal_mask,
                    position_ids=position_ids,
                    past_key_value=past_key_values,
                    output_attentions=output_attentions,
                    use_cache=use_cache,
                    cache_position=cache_position,
                )

            hidden_states = layer_outputs[0]

            if output_attentions:
                if layer_outputs[1] is not None:
                    # append attentions only of attention layers. Mamba layers return `None` as the attention weights
                    all_self_attns += (layer_outputs[1],)

        hidden_states = self.final_layernorm(hidden_states)

        # add hidden states from the last decoder layer
        if output_hidden_states:
            all_hidden_states += (hidden_states,)

        if past_key_values and not past_key_values.has_previous_state:
            past_key_values.has_previous_state = True
        if past_key_values is not None:
            past_key_values.seq_offset += hidden_states.shape[1]

        next_cache = None if not use_cache else past_key_values

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

    def _update_causal_mask(self, attention_mask, input_tensor, cache_position):
        if self.config._attn_implementation == "flash_attention_2":
            if attention_mask is not None and 0.0 in attention_mask:
                return attention_mask
            return None

        dtype, device = input_tensor.dtype, input_tensor.device
        min_dtype = torch.finfo(dtype).min
        sequence_length = input_tensor.shape[1]
        target_length = cache_position[-1] + 1

        causal_mask = torch.full((sequence_length, target_length), fill_value=min_dtype, dtype=dtype, device=device)
        if sequence_length != 1:
            causal_mask = torch.triu(causal_mask, diagonal=1)
        causal_mask *= torch.arange(target_length, device=device) > cache_position.reshape(-1, 1)
        causal_mask = causal_mask[None, None, :, :].expand(input_tensor.shape[0], 1, -1, -1)
        if attention_mask is not None:
            causal_mask = causal_mask.clone()  # copy to contiguous memory for in-place edit
            if attention_mask.dim() == 2:
                mask_length = attention_mask.shape[-1]
                padding_mask = causal_mask[..., :mask_length].eq(0.0) * attention_mask[:, None, None, :].eq(0.0)
                causal_mask[..., :mask_length] = causal_mask[..., :mask_length].masked_fill(padding_mask, min_dtype)

        if (
                self.config._attn_implementation == "sdpa"
                and attention_mask is not None
                and attention_mask.device.type == "cuda"
        ):
            # Attend to all tokens in fully masked rows in the causal_mask, for example the relevant first rows when
            # using left padding. This is required by F.scaled_dot_product_attention memory-efficient attention path.
            # Details: https://github.com/pytorch/pytorch/issues/110213
            causal_mask = AttentionMaskConverter._unmask_unattended(causal_mask, min_dtype)

        return causal_mask


class HybriDNAForCausalLM(HybriDNAPreTrainedModel, GenerationMixin):
    _tied_weights_keys = ["lm_head.weight"]

    def __init__(self, config: HybriDNAConfig):
        super().__init__(config)
        self.model = HybriDNAModel(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()

    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

    @add_start_docstrings_to_model_forward(HYBRIDNA_INPUTS_DOCSTRING)
    @replace_return_docstrings(output_type=CausalLMOutputWithPast, config_class=_CONFIG_FOR_DOC)
    def forward(
            self,
            input_ids: torch.LongTensor = None,
            attention_mask: Optional[torch.Tensor] = None,
            position_ids: Optional[torch.LongTensor] = None,
            past_key_values: Optional[HybridMambaAttentionDynamicCache] = None,
            inputs_embeds: Optional[torch.Tensor] = 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,
            cache_position: Optional[torch.LongTensor] = None,
            num_logits_to_keep: Optional[Union[int, None]] = None,
    ) -> Union[Tuple, CausalLMOutputWithPast]:
        r"""
        Args:
            labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
                Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
                config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
                (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.
            num_logits_to_keep (`int` or `None`, *optional*):
                Calculate logits for the last `num_logits_to_keep` tokens. If `None`, calculate logits for all
                `input_ids`. Only last token logits are needed for generation, and calculating them only for that token
                can save memory, which becomes pretty significant for long sequences.
        Returns:
        Example:
        ```python
        >>> from transformers import AutoTokenizer, AutoModelForCausalLM
        >>> model = AutoModelForCausalLM.from_pretrained("Mishamq/HybriDNA-300M", trust_remote_code=True)
        >>> tokenizer = AutoTokenizer.from_pretrained("Mishamq/HybriDNA-300M", trust_remote_code=True)
        >>> prompt = "ACGTACGTACGTACGT"
        >>> inputs = tokenizer(prompt, return_tensors="pt")
        >>> # Generate
        >>> generate_ids = model.generate(inputs.input_ids, max_length=30)
        >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
        ```"""

        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

        # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn)
        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,
            cache_position=cache_position,
            return_dict=return_dict,
        )

        hidden_states = outputs[0]
        if num_logits_to_keep is None:
            logits = self.lm_head(hidden_states)
        else:
            logits = self.lm_head(hidden_states[..., -num_logits_to_keep:, :])
        logits = logits.float()

        loss = None
        if labels is not None:
            # Shift so that tokens < n predict n
            shift_logits = logits[..., :-1, :].contiguous()
            shift_labels = labels[..., 1:].contiguous()
            # Flatten the tokens
            loss_fct = CrossEntropyLoss()
            shift_logits = shift_logits.view(-1, self.config.vocab_size)
            shift_labels = shift_labels.view(-1)
            # Enable model parallelism
            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,
            cache_position=None,
            **kwargs,
    ):
        # Check if we need to create our custom cache
        # This handles the case where generate() creates a default DynamicCache
        empty_past_kv = past_key_values is None or not isinstance(past_key_values, HybridMambaAttentionDynamicCache)

        # Omit tokens covered by past_key_values
        if not empty_past_kv:
            past_length = cache_position[0] if cache_position is not None else attention_mask.shape[1]
            max_cache_length = self.config.sliding_window
            # Keep only the unprocessed tokens:
            # 1 - If the length of the attention_mask exceeds the length of input_ids, then we are in a setting where
            # some of the inputs are exclusively passed as part of the cache (e.g. when passing input_embeds as
            # input)
            if attention_mask is not None and attention_mask.shape[1] > input_ids.shape[1]:
                input_ids = input_ids[:, -(attention_mask.shape[1] - past_length) :]
            # 2 - If the past_length is smaller than input_ids', then input_ids holds all input tokens. We can discard
            # input_ids based on the past_length.
            elif past_length < input_ids.shape[1]:
                input_ids = input_ids[:, past_length:]
            # 3 - Otherwise (past_length >= input_ids.shape[1]), let's assume input_ids only has unprocessed tokens.

            # If we are about to go beyond the maximum cache length, we need to crop the input attention mask.
            if (
                    max_cache_length is not None
                    and attention_mask is not None
                    and past_length + input_ids.shape[1] > max_cache_length
            ):
                attention_mask = attention_mask[:, -max_cache_length:]
        else:
            past_key_values = HybridMambaAttentionDynamicCache(
                self.config, input_ids.shape[0], self.dtype, device=self.device
            )

        position_ids = kwargs.get("position_ids", None)
        if attention_mask is not None and position_ids is None:
            # create position_ids on the fly for batch generation
            position_ids = attention_mask.long().cumsum(-1) - 1
            position_ids.masked_fill_(attention_mask == 0, 1)
            if not empty_past_kv:
                position_ids = position_ids[:, -input_ids.shape[1] :]

        # if `inputs_embeds` are passed, we only want to use them in the 1st generation step
        if inputs_embeds is not None and empty_past_kv:
            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,
                "num_logits_to_keep": self.config.num_logits_to_keep,
                "cache_position": cache_position,
            }
        )
        return model_inputs


@add_start_docstrings(
    """
    The HybriDNA Model with a sequence classification head on top (linear layer).
    [`HybriDNAForSequenceClassification`] uses the last token in order to do the classification, as other causal models
    (e.g. GPT-2) do.
    Since it does classification on the last token, it requires to know the position of the last token. If a
    `pad_token_id` is defined in the configuration, it finds the last token that is not a padding token in each row. If
    no `pad_token_id` is defined, it simply takes the last value in each row of the batch. Since it cannot guess the
    padding tokens when `inputs_embeds` are passed instead of `input_ids`, it does the same (take the last value in
    each row of the batch).
    """,
    HYBRIDNA_START_DOCSTRING,
)
class HybriDNAForSequenceClassification(HybriDNAPreTrainedModel):
    def __init__(self, config):
        super().__init__(config)
        self.num_labels = config.num_labels
        self.model = HybriDNAModel(config)
        self.score = nn.Linear(config.hidden_size, self.num_labels, bias=False)
        self.dropout = nn.Dropout(0)
        # Initialize weights and apply final processing
        self.post_init()

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

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

    @add_start_docstrings_to_model_forward(HYBRIDNA_INPUTS_DOCSTRING)
    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.Tensor]] = None,
            inputs_embeds: Optional[torch.Tensor] = 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, SequenceClassifierOutputWithPast]:
        r"""
        labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
            Labels for computing the sequence classification/regression loss. Indices should be in `[0, ...,
            config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If
            `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
        """
        return_dict = return_dict if return_dict is not None else self.config.use_return_dict

        transformer_outputs = self.model(
            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 = transformer_outputs[0]
        hidden_states = self.dropout(hidden_states)
        logits = self.score(hidden_states)

        if input_ids is not None:
            batch_size = input_ids.shape[0]
        else:
            batch_size = inputs_embeds.shape[0]

        if self.config.pad_token_id is None and batch_size != 1:
            raise ValueError("Cannot handle batch sizes > 1 if no padding token is defined.")
        if self.config.pad_token_id is None:
            sequence_lengths = -1
        else:
            if input_ids is not None:
                # if no pad token found, use modulo instead of reverse indexing for ONNX compatibility
                sequence_lengths = torch.eq(input_ids, self.config.pad_token_id).int().argmax(-1) - 1
                sequence_lengths = sequence_lengths % input_ids.shape[-1]
                sequence_lengths = sequence_lengths.to(logits.device)
            else:
                sequence_lengths = -1

        pooled_logits = logits[torch.arange(batch_size, device=logits.device), sequence_lengths]

        loss = None
        if labels is not None:
            labels = labels.to(logits.device)
            if self.config.problem_type is None:
                if self.num_labels == 1:
                    self.config.problem_type = "regression"
                elif self.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int):
                    self.config.problem_type = "single_label_classification"
                else:
                    self.config.problem_type = "multi_label_classification"

            if self.config.problem_type == "regression":
                loss_fct = MSELoss()
                if self.num_labels == 1:
                    loss = loss_fct(pooled_logits.squeeze(), labels.squeeze())
                else:
                    loss = loss_fct(pooled_logits, labels)
            elif self.config.problem_type == "single_label_classification":
                loss_fct = CrossEntropyLoss()
                loss = loss_fct(pooled_logits.view(-1, self.num_labels), labels.view(-1))
            elif self.config.problem_type == "multi_label_classification":
                loss_fct = BCEWithLogitsLoss()
                loss = loss_fct(pooled_logits, labels)
        if not return_dict:
            output = (pooled_logits,) + transformer_outputs[1:]
            return ((loss,) + output) if loss is not None else output

        return SequenceClassifierOutputWithPast(
            loss=loss,
            logits=pooled_logits,
            past_key_values=transformer_outputs.past_key_values,
            hidden_states=transformer_outputs.hidden_states,
            attentions=transformer_outputs.attentions,
        )

@add_start_docstrings(
    """
    The HybriDNA Model with a sequence classification head on top (linear layer along with RC Echo Embedding).
    The input sequence is concatenated with its reverse complement before being processed by the model.
    [`HybriDNAForSequenceClassificationRCEcho`]
    """,
    HYBRIDNA_START_DOCSTRING,
)
class HybriDNAForSequenceClassificationRCEcho(HybriDNAPreTrainedModel):
    def __init__(self, config):
        super().__init__(config)
        self.num_labels = config.num_labels
        self.model = HybriDNAModel(config)
        self.score = nn.Linear(config.hidden_size, self.num_labels, bias=False)

        # Initialize weights and apply final processing
        self.post_init()
        self.dropout = nn.Dropout(0.05)

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

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

    def _reverse_complement_tokens(self, input_ids: torch.Tensor) -> torch.Tensor:
        """
        Reverse complement DNA token IDs.
        Token mapping: A=7, C=8, G=9, T=10, N=11
        Complement: A(7)↔T(10), C(8)↔G(9), N(11)→N(11)
        Special tokens (0-6) are preserved as-is.
        """
        rc = input_ids.clone()
        # Swap A(7) ↔ T(10)
        is_A = (input_ids == 7)
        is_T = (input_ids == 10)
        rc[is_A] = 10
        rc[is_T] = 7
        # Swap C(8) ↔ G(9)
        is_C = (input_ids == 8)
        is_G = (input_ids == 9)
        rc[is_C] = 9
        rc[is_G] = 8
        # N(11) stays N, special tokens (0-6) stay as-is
        # Reverse along sequence dimension
        rc = torch.flip(rc, dims=[1])
        return rc

    @add_start_docstrings_to_model_forward(HYBRIDNA_INPUTS_DOCSTRING)
    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.Tensor]] = None,
        inputs_embeds: Optional[torch.Tensor] = 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, SequenceClassifierOutputWithPast]:
        return_dict = return_dict if return_dict is not None else self.config.use_return_dict

        # Process input_ids to ignore mask and concatenate with reverse complement
        if input_ids is not None:
            # Create a mask to identify non-masked tokens (1 where token is not masked)
            non_masked_tokens = (input_ids != self.config.pad_token_id).int()

            # Apply mask and compute reverse complement of the masked sequence
            masked_input_ids = input_ids * non_masked_tokens
            rc_input_ids = self._reverse_complement_tokens(masked_input_ids)

            # Concatenate original sequence with its reverse complement
            repeated_input_ids = torch.cat([masked_input_ids, rc_input_ids], dim=1)

            # Create a new attention mask for the repeated sequence
            if attention_mask is not None:
                repeated_attention_mask = torch.cat([attention_mask, attention_mask], dim=1)
            else:
                repeated_attention_mask = None

            input_ids = repeated_input_ids
            attention_mask = repeated_attention_mask

        # Forward pass through the model
        transformer_outputs = self.model(
            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 = transformer_outputs[0]

        # Calculate the mean of the second sequence
        sequence_length = hidden_states.shape[1] // 2
        second_half_hidden_states = hidden_states[:, sequence_length:, :]  # Select the second half
        if attention_mask is not None:
            second_half_attention_mask = attention_mask[:, sequence_length:]  # Corresponding attention mask
            sum_hidden_states = (second_half_hidden_states * second_half_attention_mask.unsqueeze(-1)).sum(dim=1)
            sum_mask = second_half_attention_mask.sum(dim=1, keepdim=True)
            mean_hidden_states = sum_hidden_states / sum_mask
        else:
            mean_hidden_states = second_half_hidden_states.mean(dim=1)

        # Apply dropout
        mean_hidden_states = self.dropout(mean_hidden_states)
        logits = self.score(mean_hidden_states)

        loss = None
        if labels is not None:
            labels = labels.to(logits.device)
            if self.config.problem_type is None:
                if self.num_labels == 1:
                    self.config.problem_type = "regression"
                elif self.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int):
                    self.config.problem_type = "single_label_classification"
                else:
                    self.config.problem_type = "multi_label_classification"

            if self.config.problem_type == "regression":
                loss_fct = MSELoss()
                if self.num_labels == 1:
                    loss = loss_fct(logits.squeeze(), labels.squeeze())
                else:
                    loss = loss_fct(logits, labels)
            elif self.config.problem_type == "single_label_classification":
                loss_fct = CrossEntropyLoss()
                loss = loss_fct(logits.view(-1, self.num_labels), labels.view(-1))
            elif self.config.problem_type == "multi_label_classification":
                loss_fct = BCEWithLogitsLoss()
                loss = loss_fct(logits, labels)
        if not return_dict:
            output = (logits,) + transformer_outputs[1:]
            return ((loss,) + output) if loss is not None else output

        return SequenceClassifierOutputWithPast(
            loss=loss,
            logits=logits,
            past_key_values=transformer_outputs.past_key_values,
            hidden_states=transformer_outputs.hidden_states,
            attentions=transformer_outputs.attentions,
        )