modeling_qwen3_vl.py

#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
#           This file was automatically generated from src/transformers/models/qwen3_vl/modular_qwen3_vl.py.
#               Do NOT edit this file manually as any edits will be overwritten by the generation of
#             the file from the modular. If any change should be done, please apply the change to the
#                          modular_qwen3_vl.py file directly. One of our CI enforces this.
#                🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨
# coding=utf-8
# Copyright 2025 The Qwen Team and The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

from dataclasses import dataclass
from typing import Any, Callable, Optional, Union, Dict
import gc
import math

import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.distributed as dist

from transformers.activations import ACT2FN
from transformers.cache_utils import Cache, DynamicCache
from transformers.generation import GenerationMixin
from transformers.integrations import use_kernel_forward_from_hub
from transformers.masking_utils import create_causal_mask
from transformers.modeling_flash_attention_utils import FlashAttentionKwargs
from transformers.modeling_layers import GradientCheckpointingLayer
from transformers.modeling_outputs import BaseModelOutputWithPast, ModelOutput
from transformers.modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update
from transformers.modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel
from transformers.processing_utils import Unpack
from transformers.utils import TransformersKwargs, auto_docstring, is_torchdynamo_compiling
from transformers.utils.deprecation import deprecate_kwarg
from transformers.utils.generic import check_model_inputs
from qwenvl.model.configuration_qwen3_vl import Qwen3VLConfig, Qwen3VLTextConfig, Qwen3VLVisionConfig
from liger_kernel.transformers.model.loss_utils import LigerForCausalLMLoss
from qwenvl.model.modeling_whisper import WhisperEncoder
from qwenvl.model.ttt.ttt_layer import TTTWrapper, SSMGating
from qwenvl.model.ttt.configs import ModelConfig as TTTModelConfig

class Qwen3VLVisionMLP(nn.Module):
    def __init__(self, config):
        super().__init__()
        self.hidden_size = config.hidden_size
        self.intermediate_size = config.intermediate_size
        self.linear_fc1 = nn.Linear(self.hidden_size, self.intermediate_size, bias=True)
        self.linear_fc2 = nn.Linear(self.intermediate_size, self.hidden_size, bias=True)
        self.act_fn = ACT2FN[config.hidden_act]

    def forward(self, hidden_state):
        return self.linear_fc2(self.act_fn(self.linear_fc1(hidden_state)))


class Qwen3VLVisionPatchEmbed(nn.Module):
    def __init__(self, config) -> None:
        super().__init__()
        self.patch_size = config.patch_size
        self.temporal_patch_size = config.temporal_patch_size
        self.in_channels = config.in_channels
        self.embed_dim = config.hidden_size

        kernel_size = [self.temporal_patch_size, self.patch_size, self.patch_size]
        self.proj = nn.Conv3d(self.in_channels, self.embed_dim, kernel_size=kernel_size, stride=kernel_size, bias=True)

    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
        target_dtype = self.proj.weight.dtype
        hidden_states = hidden_states.view(
            -1, self.in_channels, self.temporal_patch_size, self.patch_size, self.patch_size
        )
        hidden_states = self.proj(hidden_states.to(dtype=target_dtype)).view(-1, self.embed_dim)
        return hidden_states


class Qwen3VLVisionRotaryEmbedding(nn.Module):
    inv_freq: torch.Tensor  # fix linting for `register_buffer`

    def __init__(self, dim: int, theta: float = 10000.0) -> None:
        super().__init__()
        inv_freq = 1.0 / (theta ** (torch.arange(0, dim, 2, dtype=torch.float) / dim))
        self.register_buffer("inv_freq", inv_freq, persistent=False)

    def forward(self, seqlen: int) -> torch.Tensor:
        seq = torch.arange(seqlen, device=self.inv_freq.device, dtype=self.inv_freq.dtype)
        freqs = torch.outer(seq, self.inv_freq)
        return freqs


class Qwen3VLVisionPatchMerger(nn.Module):
    def __init__(self, config: Qwen3VLVisionConfig, use_postshuffle_norm=False) -> None:
        super().__init__()
        self.hidden_size = config.hidden_size * (config.spatial_merge_size**2)
        self.use_postshuffle_norm = use_postshuffle_norm
        self.norm = nn.LayerNorm(self.hidden_size if use_postshuffle_norm else config.hidden_size, eps=1e-6)
        self.linear_fc1 = nn.Linear(self.hidden_size, self.hidden_size)
        self.act_fn = nn.GELU()
        self.linear_fc2 = nn.Linear(self.hidden_size, config.out_hidden_size)

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        x = self.norm(x.view(-1, self.hidden_size) if self.use_postshuffle_norm else x).view(-1, self.hidden_size)
        x = self.linear_fc2(self.act_fn(self.linear_fc1(x)))
        return x


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_vision(
    q: torch.Tensor, k: torch.Tensor, cos: torch.Tensor, sin: torch.Tensor
) -> tuple[torch.Tensor, torch.Tensor]:
    orig_q_dtype = q.dtype
    orig_k_dtype = k.dtype
    q, k = q.float(), k.float()
    cos, sin = cos.unsqueeze(-2).float(), sin.unsqueeze(-2).float()
    q_embed = (q * cos) + (rotate_half(q) * sin)
    k_embed = (k * cos) + (rotate_half(k) * sin)
    q_embed = q_embed.to(orig_q_dtype)
    k_embed = k_embed.to(orig_k_dtype)
    return q_embed, k_embed


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


def eager_attention_forward(
    module: nn.Module,
    query: torch.Tensor,
    key: torch.Tensor,
    value: torch.Tensor,
    attention_mask: Optional[torch.Tensor],
    scaling: float,
    dropout: float = 0.0,
    **kwargs: Unpack[TransformersKwargs],
):
    key_states = repeat_kv(key, module.num_key_value_groups)
    value_states = repeat_kv(value, module.num_key_value_groups)

    attn_weights = torch.matmul(query, key_states.transpose(2, 3)) * scaling
    if attention_mask is not None:
        causal_mask = attention_mask[:, :, :, : key_states.shape[-2]]
        attn_weights = attn_weights + causal_mask

    attn_weights = nn.functional.softmax(attn_weights, dim=-1, dtype=torch.float32).to(query.dtype)
    attn_weights = nn.functional.dropout(attn_weights, p=dropout, training=module.training)
    attn_output = torch.matmul(attn_weights, value_states)
    attn_output = attn_output.transpose(1, 2).contiguous()

    return attn_output, attn_weights


class Qwen3VLVisionAttention(nn.Module):
    def __init__(self, config: Qwen3VLVisionConfig) -> None:
        super().__init__()
        self.dim = config.hidden_size
        self.num_heads = config.num_heads
        self.head_dim = self.dim // self.num_heads
        self.num_key_value_groups = 1  # needed for eager attention
        self.qkv = nn.Linear(self.dim, self.dim * 3, bias=True)
        self.proj = nn.Linear(self.dim, self.dim)
        self.scaling = self.head_dim**-0.5
        self.config = config
        self.attention_dropout = 0.0
        self.is_causal = False

    def forward(
        self,
        hidden_states: torch.Tensor,
        cu_seqlens: torch.Tensor,
        rotary_pos_emb: Optional[torch.Tensor] = None,
        position_embeddings: Optional[tuple[torch.Tensor, torch.Tensor]] = None,
        **kwargs,
    ) -> torch.Tensor:
        seq_length = hidden_states.shape[0]
        query_states, key_states, value_states = (
            self.qkv(hidden_states).reshape(seq_length, 3, self.num_heads, -1).permute(1, 0, 2, 3).unbind(0)
        )
        cos, sin = position_embeddings
        query_states, key_states = apply_rotary_pos_emb_vision(query_states, key_states, cos, sin)

        query_states = query_states.transpose(0, 1).unsqueeze(0)
        key_states = key_states.transpose(0, 1).unsqueeze(0)
        value_states = value_states.transpose(0, 1).unsqueeze(0)

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

        if self.config._attn_implementation == "flash_attention_2":
            # Flash Attention 2: Use cu_seqlens for variable length attention
            max_seqlen = (cu_seqlens[1:] - cu_seqlens[:-1]).max()
            attn_output, _ = attention_interface(
                self,
                query_states,
                key_states,
                value_states,
                attention_mask=None,
                scaling=self.scaling,
                dropout=0.0 if not self.training else self.attention_dropout,
                cu_seq_lens_q=cu_seqlens,
                cu_seq_lens_k=cu_seqlens,
                max_length_q=max_seqlen,
                max_length_k=max_seqlen,
                is_causal=False,
                **kwargs,
            )
        else:
            # Other implementations: Process each chunk separately
            lengths = cu_seqlens[1:] - cu_seqlens[:-1]
            splits = [
                torch.split(tensor, lengths.tolist(), dim=2) for tensor in (query_states, key_states, value_states)
            ]

            attn_outputs = [
                attention_interface(
                    self,
                    q,
                    k,
                    v,
                    attention_mask=None,
                    scaling=self.scaling,
                    dropout=0.0 if not self.training else self.attention_dropout,
                    is_causal=False,
                    **kwargs,
                )[0]
                for q, k, v in zip(*splits)
            ]
            attn_output = torch.cat(attn_outputs, dim=1)

        attn_output = attn_output.reshape(seq_length, -1).contiguous()
        attn_output = self.proj(attn_output)
        return attn_output


class Qwen3VLVisionBlock(GradientCheckpointingLayer):
    def __init__(self, config, attn_implementation: str = "sdpa") -> None:
        super().__init__()
        self.norm1 = nn.LayerNorm(config.hidden_size, eps=1e-6)
        self.norm2 = nn.LayerNorm(config.hidden_size, eps=1e-6)
        self.attn = Qwen3VLVisionAttention(config=config)
        self.mlp = Qwen3VLVisionMLP(config=config)

    def forward(
        self,
        hidden_states: torch.Tensor,
        cu_seqlens: torch.Tensor,
        rotary_pos_emb: Optional[torch.Tensor] = None,
        position_embeddings: Optional[tuple[torch.Tensor, torch.Tensor]] = None,
        **kwargs,
    ) -> torch.Tensor:
        hidden_states = hidden_states + self.attn(
            self.norm1(hidden_states),
            cu_seqlens=cu_seqlens,
            rotary_pos_emb=rotary_pos_emb,
            position_embeddings=position_embeddings,
            **kwargs,
        )
        hidden_states = hidden_states + self.mlp(self.norm2(hidden_states))
        return hidden_states


class Qwen3VLTextRotaryEmbedding(nn.Module):
    inv_freq: torch.Tensor  # fix linting for `register_buffer`

    def __init__(self, config: Qwen3VLTextConfig, device=None):
        super().__init__()
        if hasattr(config, "rope_scaling") and config.rope_scaling is not None:
            self.rope_type = config.rope_scaling.get("rope_type", "default")
        else:
            self.rope_type = "default"
        self.max_seq_len_cached = config.max_position_embeddings
        self.original_max_seq_len = config.max_position_embeddings

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

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

        self.mrope_section = config.rope_scaling.get("mrope_section", [24, 20, 20])

    def apply_interleaved_mrope(self, freqs, mrope_section):
        """Apply interleaved MRoPE to 3D rotary embeddings.
        Reorganizes frequency layout from chunked [TTT...HHH...WWW] to
        interleaved [THTHWHTHW...TT], preserving frequency continuity.
        args:
            x: (3, bs, seq_len, head_dim // 2)
            mrope_section: (3,)
        returns:
            x_t: (bs, seq_len, head_dim // 2)
        """
        freqs_t = freqs[0]  # just overwrite the first dimension T
        for dim, offset in enumerate((1, 2), start=1):  # H, W
            length = mrope_section[dim] * 3
            idx = slice(offset, length, 3)
            freqs_t[..., idx] = freqs[dim, ..., idx]
        return freqs_t

    @torch.no_grad()
    @dynamic_rope_update  # power user: used with advanced RoPE types (e.g. dynamic rope)
    def forward(self, x, position_ids):
        # In contrast to other models, Qwen3VL has different position ids for the grids
        # So we expand the inv_freq to shape (3, ...)
        if position_ids.ndim == 2:
            position_ids = position_ids[None, ...].expand(3, position_ids.shape[0], -1)
        inv_freq_expanded = self.inv_freq[None, None, :, None].float().expand(3, position_ids.shape[1], -1, 1)
        position_ids_expanded = position_ids[:, :, None, :].float()  # shape (3, bs, 1, positions)

        device_type = x.device.type if isinstance(x.device.type, str) and x.device.type != "mps" else "cpu"
        with torch.autocast(device_type=device_type, enabled=False):  # Force float32
            freqs = (inv_freq_expanded.float() @ position_ids_expanded.float()).transpose(2, 3)
            freqs = self.apply_interleaved_mrope(freqs, self.mrope_section)
            emb = torch.cat((freqs, freqs), dim=-1)
            cos = emb.cos() * self.attention_scaling
            sin = emb.sin() * self.attention_scaling

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


@use_kernel_forward_from_hub("RMSNorm")
class Qwen3VLTextRMSNorm(nn.Module):
    def __init__(self, hidden_size, eps: float = 1e-6) -> None:
        """
        Qwen3VLTextRMSNorm is equivalent to T5LayerNorm
        """
        super().__init__()
        self.weight = nn.Parameter(torch.ones(hidden_size))
        self.variance_epsilon = eps

    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
        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)

    def extra_repr(self):
        return f"{tuple(self.weight.shape)}, eps={self.variance_epsilon}"


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

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


class Qwen3VLTextAttention(nn.Module):
    """Multi-headed attention from 'Attention Is All You Need' paper"""

    def __init__(self, config: Qwen3VLTextConfig, layer_idx: int):
        super().__init__()
        self.config = config
        self.layer_idx = layer_idx
        self.head_dim = getattr(config, "head_dim", config.hidden_size // config.num_attention_heads)
        self.num_key_value_groups = config.num_attention_heads // config.num_key_value_heads
        self.scaling = self.head_dim**-0.5
        self.attention_dropout = config.attention_dropout
        self.is_causal = True

        self.q_proj = nn.Linear(
            config.hidden_size, config.num_attention_heads * self.head_dim, bias=config.attention_bias
        )
        self.k_proj = nn.Linear(
            config.hidden_size, config.num_key_value_heads * self.head_dim, bias=config.attention_bias
        )
        self.v_proj = nn.Linear(
            config.hidden_size, config.num_key_value_heads * self.head_dim, bias=config.attention_bias
        )
        self.o_proj = nn.Linear(
            config.num_attention_heads * self.head_dim, config.hidden_size, bias=config.attention_bias
        )
        self.q_norm = Qwen3VLTextRMSNorm(self.head_dim, eps=config.rms_norm_eps)  # unlike olmo, only on the head dim!
        self.k_norm = Qwen3VLTextRMSNorm(
            self.head_dim, eps=config.rms_norm_eps
        )  # thus post q_norm does not need reshape

    @deprecate_kwarg("past_key_value", new_name="past_key_values", version="4.58")
    def forward(
        self,
        hidden_states: torch.Tensor,
        position_embeddings: tuple[torch.Tensor, torch.Tensor],
        attention_mask: Optional[torch.Tensor],
        past_key_values: Optional[Cache] = None,
        cache_position: Optional[torch.LongTensor] = None,
        **kwargs: Unpack[FlashAttentionKwargs],
    ) -> tuple[torch.Tensor, Optional[torch.Tensor]]:
        input_shape = hidden_states.shape[:-1]
        hidden_shape = (*input_shape, -1, self.head_dim)

        query_states = self.q_norm(self.q_proj(hidden_states).view(hidden_shape)).transpose(1, 2)
        key_states = self.k_norm(self.k_proj(hidden_states).view(hidden_shape)).transpose(1, 2)
        value_states = self.v_proj(hidden_states).view(hidden_shape).transpose(1, 2)

        cos, sin = position_embeddings
        query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin)

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

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

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

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


class Qwen3VLTextMLP(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):
        down_proj = self.down_proj(self.act_fn(self.gate_proj(x)) * self.up_proj(x))
        return down_proj


class Qwen3VLTextDecoderLayer(GradientCheckpointingLayer):
    def __init__(self, config: Qwen3VLTextConfig, layer_idx: int):
        super().__init__()
        self.hidden_size = config.hidden_size

        self.self_attn = Qwen3VLTextAttention(config=config, layer_idx=layer_idx)

        self.mlp = Qwen3VLTextMLP(config)
        self.input_layernorm = Qwen3VLTextRMSNorm(config.hidden_size, eps=config.rms_norm_eps)
        self.post_attention_layernorm = Qwen3VLTextRMSNorm(config.hidden_size, eps=config.rms_norm_eps)

    @deprecate_kwarg("past_key_value", new_name="past_key_values", version="4.58")
    def forward(
        self,
        hidden_states: torch.Tensor,
        position_embeddings: tuple[torch.Tensor, torch.Tensor],
        attention_mask: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.LongTensor] = None,
        past_key_values: Optional[Cache] = None,
        use_cache: Optional[bool] = False,
        cache_position: Optional[torch.LongTensor] = None,
        **kwargs: Unpack[TransformersKwargs],
    ) -> torch.Tensor:
        residual = hidden_states
        hidden_states = self.input_layernorm(hidden_states)
        # Self Attention
        hidden_states, _ = self.self_attn(
            hidden_states=hidden_states,
            attention_mask=attention_mask,
            position_ids=position_ids,
            past_key_values=past_key_values,
            use_cache=use_cache,
            cache_position=cache_position,
            position_embeddings=position_embeddings,
            **kwargs,
        )
        hidden_states = residual + hidden_states

        # Fully Connected
        residual = hidden_states
        hidden_states = self.post_attention_layernorm(hidden_states)
        hidden_states = self.mlp(hidden_states)
        hidden_states = residual + hidden_states
        return hidden_states


@dataclass
@auto_docstring(
    custom_intro="""
    Base class for Llava outputs, with hidden states and attentions.
    """
)
class Qwen3VLModelOutputWithPast(ModelOutput):
    r"""
    past_key_values (`Cache`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
        It is a [`~cache_utils.Cache`] instance. For more details, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache).

        Contains pre-computed hidden-states (key and values in the self-attention blocks) that can be used (see
        `past_key_values` input) to speed up sequential decoding.
    rope_deltas (`torch.LongTensor` of shape `(batch_size, )`, *optional*):
        The rope index difference between sequence length and multimodal rope.
    """

    last_hidden_state: Optional[torch.FloatTensor] = None
    past_key_values: Optional[Cache] = None
    hidden_states: Optional[tuple[torch.FloatTensor]] = None
    attentions: Optional[tuple[torch.FloatTensor]] = None
    rope_deltas: Optional[torch.LongTensor] = None
    labels: Optional[torch.LongTensor] = None
    memory_triplets: Optional[dict] = None


@auto_docstring
class Qwen3VLPreTrainedModel(PreTrainedModel):
    config: Qwen3VLConfig
    base_model_prefix = "model"
    supports_gradient_checkpointing = True
    _no_split_modules = ["Qwen3VLTextDecoderLayer", "Qwen3VLVisionBlock"]
    _skip_keys_device_placement = "past_key_values"
    _supports_flash_attn = True
    _supports_sdpa = True

    _can_compile_fullgraph = True
    _supports_attention_backend = True
    _can_record_outputs = {
        "hidden_states": Qwen3VLTextDecoderLayer,
        "attentions": Qwen3VLTextAttention,
    }


class Qwen3VLVisionModel(Qwen3VLPreTrainedModel):
    config: Qwen3VLVisionConfig
    _no_split_modules = ["Qwen3VLVisionBlock"]

    def __init__(self, config, *inputs, **kwargs) -> None:
        super().__init__(config, *inputs, **kwargs)
        self.spatial_merge_size = config.spatial_merge_size
        self.patch_size = config.patch_size
        self.spatial_merge_unit = self.spatial_merge_size * self.spatial_merge_size

        self.patch_embed = Qwen3VLVisionPatchEmbed(
            config=config,
        )

        self.pos_embed = nn.Embedding(config.num_position_embeddings, config.hidden_size)
        self.num_grid_per_side = int(config.num_position_embeddings**0.5)

        head_dim = config.hidden_size // config.num_heads
        self.rotary_pos_emb = Qwen3VLVisionRotaryEmbedding(head_dim // 2)

        self.blocks = nn.ModuleList([Qwen3VLVisionBlock(config) for _ in range(config.depth)])
        self.merger = Qwen3VLVisionPatchMerger(
            config=config,
            use_postshuffle_norm=False,
        )

        self.deepstack_visual_indexes = config.deepstack_visual_indexes
        self.deepstack_merger_list = nn.ModuleList(
            [
                Qwen3VLVisionPatchMerger(
                    config=config,
                    use_postshuffle_norm=True,
                )
                for _ in range(len(config.deepstack_visual_indexes))
            ]
        )

        self.gradient_checkpointing = False

    def rot_pos_emb(self, grid_thw: torch.Tensor) -> torch.Tensor:
        merge_size = self.spatial_merge_size

        max_hw = int(grid_thw[:, 1:].max().item())
        freq_table = self.rotary_pos_emb(max_hw)  # (max_hw, dim // 2)
        device = freq_table.device

        total_tokens = int(torch.prod(grid_thw, dim=1).sum().item())
        pos_ids = torch.empty((total_tokens, 2), dtype=torch.long, device=device)

        offset = 0
        for num_frames, height, width in grid_thw:
            merged_h, merged_w = height // merge_size, width // merge_size

            block_rows = torch.arange(merged_h, device=device)  # block row indices
            block_cols = torch.arange(merged_w, device=device)  # block col indices
            intra_row = torch.arange(merge_size, device=device)  # intra-block row offsets
            intra_col = torch.arange(merge_size, device=device)  # intra-block col offsets

            # Compute full-resolution positions
            row_idx = block_rows[:, None, None, None] * merge_size + intra_row[None, None, :, None]
            col_idx = block_cols[None, :, None, None] * merge_size + intra_col[None, None, None, :]

            row_idx = row_idx.expand(merged_h, merged_w, merge_size, merge_size).reshape(-1)
            col_idx = col_idx.expand(merged_h, merged_w, merge_size, merge_size).reshape(-1)

            coords = torch.stack((row_idx, col_idx), dim=-1)

            if num_frames > 1:
                coords = coords.repeat(num_frames, 1)

            num_tokens = coords.shape[0]
            pos_ids[offset : offset + num_tokens] = coords
            offset += num_tokens

        embeddings = freq_table[pos_ids]  # lookup rotary embeddings
        embeddings = embeddings.flatten(1)
        return embeddings

    def fast_pos_embed_interpolate(self, grid_thw):
        grid_ts, grid_hs, grid_ws = grid_thw[:, 0], grid_thw[:, 1], grid_thw[:, 2]

        idx_list = [[] for _ in range(4)]
        weight_list = [[] for _ in range(4)]

        for t, h, w in zip(grid_ts, grid_hs, grid_ws):
            h_idxs = torch.linspace(0, self.num_grid_per_side - 1, h)
            w_idxs = torch.linspace(0, self.num_grid_per_side - 1, w)

            h_idxs_floor = h_idxs.int()
            w_idxs_floor = w_idxs.int()
            h_idxs_ceil = (h_idxs.int() + 1).clip(max=self.num_grid_per_side - 1)
            w_idxs_ceil = (w_idxs.int() + 1).clip(max=self.num_grid_per_side - 1)

            dh = h_idxs - h_idxs_floor
            dw = w_idxs - w_idxs_floor

            base_h = h_idxs_floor * self.num_grid_per_side
            base_h_ceil = h_idxs_ceil * self.num_grid_per_side

            indices = [
                (base_h[None].T + w_idxs_floor[None]).flatten(),
                (base_h[None].T + w_idxs_ceil[None]).flatten(),
                (base_h_ceil[None].T + w_idxs_floor[None]).flatten(),
                (base_h_ceil[None].T + w_idxs_ceil[None]).flatten(),
            ]

            weights = [
                ((1 - dh)[None].T * (1 - dw)[None]).flatten(),
                ((1 - dh)[None].T * dw[None]).flatten(),
                (dh[None].T * (1 - dw)[None]).flatten(),
                (dh[None].T * dw[None]).flatten(),
            ]

            for i in range(4):
                idx_list[i].extend(indices[i].tolist())
                weight_list[i].extend(weights[i].tolist())

        idx_tensor = torch.tensor(idx_list, dtype=torch.long, device=self.pos_embed.weight.device)
        weight_tensor = torch.tensor(
            weight_list, dtype=self.pos_embed.weight.dtype, device=self.pos_embed.weight.device
        )
        pos_embeds = self.pos_embed(idx_tensor) * weight_tensor[:, :, None]
        patch_pos_embeds = pos_embeds[0] + pos_embeds[1] + pos_embeds[2] + pos_embeds[3]

        patch_pos_embeds = patch_pos_embeds.split([h * w for h, w in zip(grid_hs, grid_ws)])

        patch_pos_embeds_permute = []
        merge_size = self.config.spatial_merge_size
        for pos_embed, t, h, w in zip(patch_pos_embeds, grid_ts, grid_hs, grid_ws):
            pos_embed = pos_embed.repeat(t, 1)
            pos_embed = (
                pos_embed.view(t, h // merge_size, merge_size, w // merge_size, merge_size, -1)
                .permute(0, 1, 3, 2, 4, 5)
                .flatten(0, 4)
            )
            patch_pos_embeds_permute.append(pos_embed)
        patch_pos_embeds = torch.cat(patch_pos_embeds_permute)
        return patch_pos_embeds

    def forward(self, hidden_states: torch.Tensor, grid_thw: torch.Tensor, **kwargs) -> torch.Tensor:
        """
        Args:
            hidden_states (`torch.Tensor` of shape `(seq_len, hidden_size)`):
                The final hidden states of the model.
            grid_thw (`torch.Tensor` of shape `(num_images_or_videos, 3)`):
                The temporal, height and width of feature shape of each image in LLM.

        Returns:
            `torch.Tensor`: hidden_states.
        """
        hidden_states = self.patch_embed(hidden_states)

        pos_embeds = self.fast_pos_embed_interpolate(grid_thw)
        hidden_states = hidden_states + pos_embeds

        rotary_pos_emb = self.rot_pos_emb(grid_thw)

        seq_len, _ = hidden_states.size()
        hidden_states = hidden_states.reshape(seq_len, -1)
        rotary_pos_emb = rotary_pos_emb.reshape(seq_len, -1)
        emb = torch.cat((rotary_pos_emb, rotary_pos_emb), dim=-1)
        position_embeddings = (emb.cos(), emb.sin())

        cu_seqlens = torch.repeat_interleave(grid_thw[:, 1] * grid_thw[:, 2], grid_thw[:, 0]).cumsum(
            dim=0,
            # Select dtype based on the following factors:
            #  - FA2 requires that cu_seqlens_q must have dtype int32
            #  - torch.onnx.export requires that cu_seqlens_q must have same dtype as grid_thw
            # See https://github.com/huggingface/transformers/pull/34852 for more information
            dtype=grid_thw.dtype if torch.jit.is_tracing() else torch.int32,
        )
        cu_seqlens = F.pad(cu_seqlens, (1, 0), value=0)

        deepstack_feature_lists = []
        for layer_num, blk in enumerate(self.blocks):
            hidden_states = blk(
                hidden_states,
                cu_seqlens=cu_seqlens,
                position_embeddings=position_embeddings,
                **kwargs,
            )
            if layer_num in self.deepstack_visual_indexes:
                deepstack_feature = self.deepstack_merger_list[self.deepstack_visual_indexes.index(layer_num)](
                    hidden_states
                )
                deepstack_feature_lists.append(deepstack_feature)

        hidden_states = self.merger(hidden_states)

        return hidden_states, deepstack_feature_lists


@auto_docstring(
    custom_intro=(
        "Text part of Qwen3VL, "
        "not a pure text-only model, as DeepStack integrates visual features into the early hidden states."
    )
)
class Qwen3VLTextModel(Qwen3VLPreTrainedModel):
    config: Qwen3VLTextConfig
    _no_split_modules = ["Qwen3VLTextDecoderLayer"]

    def __init__(self, config: Qwen3VLTextConfig):
        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(
            [Qwen3VLTextDecoderLayer(config, layer_idx) for layer_idx in range(config.num_hidden_layers)]
        )
        self.norm = Qwen3VLTextRMSNorm(config.hidden_size, eps=config.rms_norm_eps)
        self.rotary_emb = Qwen3VLTextRotaryEmbedding(config=config)
        self.gradient_checkpointing = False
        self.search_type = "none"
        self.workingmemsize = 0

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

    @check_model_inputs
    @auto_docstring
    def forward(
        self,
        input_ids: Optional[torch.LongTensor] = None,
        attention_mask: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.LongTensor] = None,
        past_key_values: Optional[Cache] = None,
        inputs_embeds: Optional[torch.FloatTensor] = None,
        use_cache: Optional[bool] = None,
        cache_position: Optional[torch.LongTensor] = None,
        # args for deepstack
        visual_pos_masks: Optional[torch.Tensor] = None,
        deepstack_visual_embeds: Optional[list[torch.Tensor]] = None,
        **kwargs: Unpack[FlashAttentionKwargs],
    ) -> Union[tuple, BaseModelOutputWithPast]:
        r"""
        visual_pos_masks (`torch.Tensor` of shape `(batch_size, seqlen)`, *optional*):
            The mask of the visual positions.
        deepstack_visual_embeds (`list[torch.Tensor]`, *optional*):
            The deepstack visual embeddings. The shape is (num_layers, visual_seqlen, embed_dim).
            The feature is extracted from the different visual encoder layers, and fed to the decoder
            hidden states. It's from the paper DeepStack(https://arxiv.org/abs/2406.04334).
        """
        if (input_ids is None) ^ (inputs_embeds is not None):
            raise ValueError("You must specify exactly one of input_ids or inputs_embeds")

        # torch.jit.trace() doesn't support cache objects in the output
        if use_cache and past_key_values is None and not torch.jit.is_tracing():
            past_key_values = DynamicCache(config=self.config)

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

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

        # the hard coded `3` is for temporal, height and width.
        if position_ids is None:
            position_ids = cache_position.view(1, 1, -1).expand(3, inputs_embeds.shape[0], -1)
        elif position_ids.ndim == 2:
            position_ids = position_ids[None, ...].expand(3, position_ids.shape[0], -1)

        if position_ids.ndim == 3 and position_ids.shape[0] == 4:
            text_position_ids = position_ids[0]
            position_ids = position_ids[1:]
        else:
            text_position_ids = position_ids[0]

        attention_mask = create_causal_mask(
            config=self.config,
            input_embeds=inputs_embeds,
            attention_mask=attention_mask,
            cache_position=cache_position,
            past_key_values=past_key_values,
            position_ids=text_position_ids,
        )

        hidden_states = inputs_embeds

        # create position embeddings to be shared across the decoder layers
        position_embeddings = self.rotary_emb(hidden_states, position_ids)

        # decoder layers
        for layer_idx, decoder_layer in enumerate(self.layers):
            layer_outputs = decoder_layer(
                hidden_states,
                attention_mask=attention_mask,
                position_ids=text_position_ids,
                past_key_values=past_key_values,
                cache_position=cache_position,
                position_embeddings=position_embeddings,
                **kwargs,
            )
            hidden_states = layer_outputs

            # add visual features to the hidden states of first several layers
            if deepstack_visual_embeds is not None and layer_idx in range(len(deepstack_visual_embeds)):
                hidden_states = self._deepstack_process(
                    hidden_states,
                    visual_pos_masks,
                    deepstack_visual_embeds[layer_idx],
                )

        hidden_states = self.norm(hidden_states)

        return BaseModelOutputWithPast(
            last_hidden_state=hidden_states,
            past_key_values=past_key_values,
        )

    def _deepstack_process(
        self, hidden_states: torch.Tensor, visual_pos_masks: torch.Tensor, visual_embeds: torch.Tensor
    ):
        visual_pos_masks = visual_pos_masks.to(hidden_states.device)
        visual_embeds = visual_embeds.to(hidden_states.device, hidden_states.dtype)
        local_this = hidden_states[visual_pos_masks, :].clone() + visual_embeds
        hidden_states[visual_pos_masks, :] = local_this
        return hidden_states


@auto_docstring
class Qwen3VLModel(Qwen3VLPreTrainedModel):
    base_model_prefix = ""
    _checkpoint_conversion_mapping = {}
    # Reference: fix gemma3 grad acc #37208
    accepts_loss_kwargs = False
    config: Qwen3VLConfig
    _no_split_modules = ["Qwen3VLTextDecoderLayer", "Qwen3VLVisionBlock"]

    def __init__(self, config):
        super().__init__(config)
        self.visual = Qwen3VLVisionModel._from_config(config.vision_config)
        self.config = config
        self.audio = WhisperEncoder._from_config(
            config.audio_config, attn_implementation=config._attn_implementation
        )
        self.language_model = Qwen3VLTextModel._from_config(config.text_config)
        self.rope_deltas = None  # cache rope_deltas here
        self.fixed_memory_size = 0
        self.fixed_memory_size_audio = 0
        self.stepsize = 0
        self.ttt_type = "simsample"
        self.search_type = "none"

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

    def init_ttt_layers(
        self,
        num_heads=8,
        ttt_gating=True,
        ttt_minibatch_size=1024,
        ttt_cg_type="ttt_mlp_cg",
        CG_max_iter=0,
        ttt_hidden_size=4,
        ttt_base_lr=0.1,
        freeze_ttt=False,
        memgroupsize=0,
        workingmemsize=0,
    ):
        hidden_size = self.config.text_config.hidden_size
        total_embed_num = len(self.config.vision_config.deepstack_visual_indexes) + 1
        # model_dim = hidden_size * total_embed_num
        self.ttt_configs = TTTModelConfig(
            model_dim=hidden_size,
            num_heads=num_heads,
            num_layers=1,
            ttt_base_lr=ttt_base_lr,
            mini_batch_size=ttt_minibatch_size,
            ssm_layer=ttt_cg_type,
            ttt_hidden_size=ttt_hidden_size,
        )
        self.ttt_minibatch_size = ttt_minibatch_size
        self.ttt_layers = TTTWrapper(self.ttt_configs, CG_max_iter=CG_max_iter)
        self.ttt_use_gating = ttt_gating
        if ttt_gating:
            self.ttt_gating = SSMGating(hidden_size)
        self.use_ttt = True
        self.freeze_ttt = freeze_ttt
        self.memgroupsize = memgroupsize

    def init_mem_search(self, search_type, workingmemsize=0):
        self.hidden_size = self.config.text_config.hidden_size
        if "attn" in search_type:
            self.search_query = nn.Parameter(torch.randn(self.hidden_size))
        self.search_alpha = nn.Parameter(torch.zeros(1))
        self.search_type = search_type
        self.workingmemsize = workingmemsize
        if "kvcache" in self.search_type:
            self.language_model.search_type = self.search_type
            self.language_model.workingmemsize = self.workingmemsize

    def get_input_embeddings(self):
        return self.language_model.get_input_embeddings()

    def set_input_embeddings(self, value):
        self.language_model.set_input_embeddings(value)

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

    def get_decoder(self):
        return self.language_model

    def get_rope_index(
        self,
        input_ids: Optional[torch.LongTensor] = None,
        image_grid_thw: Optional[torch.LongTensor] = None,
        video_grid_thw: Optional[torch.LongTensor] = None,
        attention_mask: Optional[torch.Tensor] = None,
    ) -> tuple[torch.Tensor, torch.Tensor]:
        """Different from the original implementation, Qwen3VL use timestamps rather than absolute time position ids."""

        # Since we use timestamps to seperate videos, like <t1> <vision_start> <frame1> <vision_end> <t2> <vision_start> <frame2> <vision_end>, the video_grid_thw should also be split
        if video_grid_thw is not None:
            video_grid_thw = torch.repeat_interleave(video_grid_thw, video_grid_thw[:, 0], dim=0)
            video_grid_thw[:, 0] = 1

        spatial_merge_size = self.config.vision_config.spatial_merge_size
        image_token_id = self.config.image_token_id
        video_token_id = self.config.video_token_id
        vision_start_token_id = self.config.vision_start_token_id
        mrope_position_deltas = []
        if input_ids is not None and (image_grid_thw is not None or video_grid_thw is not None):
            total_input_ids = input_ids
            if attention_mask is None:
                attention_mask = torch.ones_like(total_input_ids)
            position_ids = torch.ones(
                3,
                input_ids.shape[0],
                input_ids.shape[1],
                dtype=input_ids.dtype,
                device=input_ids.device,
            )
            image_index, video_index = 0, 0
            attention_mask = attention_mask.to(total_input_ids.device)
            for i, input_ids in enumerate(total_input_ids):
                input_ids = input_ids[attention_mask[i] == 1]
                image_nums, video_nums = 0, 0
                vision_start_indices = torch.argwhere(input_ids == vision_start_token_id).squeeze(1)
                vision_tokens = input_ids[vision_start_indices + 1]
                image_nums = (vision_tokens == image_token_id).sum()
                video_nums = (vision_tokens == video_token_id).sum()
                input_tokens = input_ids.tolist()
                llm_pos_ids_list: list = []
                st = 0
                remain_images, remain_videos = image_nums, video_nums
                for _ in range(image_nums + video_nums):
                    if image_token_id in input_tokens and remain_images > 0:
                        ed_image = input_tokens.index(image_token_id, st)
                    else:
                        ed_image = len(input_tokens) + 1
                    if video_token_id in input_tokens and remain_videos > 0:
                        ed_video = input_tokens.index(video_token_id, st)
                    else:
                        ed_video = len(input_tokens) + 1
                    if ed_image < ed_video:
                        t, h, w = (
                            image_grid_thw[image_index][0],
                            image_grid_thw[image_index][1],
                            image_grid_thw[image_index][2],
                        )
                        image_index += 1
                        remain_images -= 1
                        ed = ed_image

                    else:
                        t, h, w = (
                            video_grid_thw[video_index][0],
                            video_grid_thw[video_index][1],
                            video_grid_thw[video_index][2],
                        )
                        video_index += 1
                        remain_videos -= 1
                        ed = ed_video
                    llm_grid_t, llm_grid_h, llm_grid_w = (
                        t.item(),
                        h.item() // spatial_merge_size,
                        w.item() // spatial_merge_size,
                    )
                    text_len = ed - st

                    st_idx = llm_pos_ids_list[-1].max() + 1 if len(llm_pos_ids_list) > 0 else 0
                    llm_pos_ids_list.append(torch.arange(text_len).view(1, -1).expand(3, -1) + st_idx)

                    # t_index is always 0 because llm_grid_t is always 1 (we use timestamps to encode the temporal information for videos)
                    t_index = torch.arange(llm_grid_t).view(-1, 1).expand(-1, llm_grid_h * llm_grid_w).flatten()
                    h_index = torch.arange(llm_grid_h).view(1, -1, 1).expand(llm_grid_t, -1, llm_grid_w).flatten()
                    w_index = torch.arange(llm_grid_w).view(1, 1, -1).expand(llm_grid_t, llm_grid_h, -1).flatten()
                    llm_pos_ids_list.append(torch.stack([t_index, h_index, w_index]) + text_len + st_idx)
                    st = ed + llm_grid_t * llm_grid_h * llm_grid_w

                if st < len(input_tokens):
                    st_idx = llm_pos_ids_list[-1].max() + 1 if len(llm_pos_ids_list) > 0 else 0
                    text_len = len(input_tokens) - st
                    llm_pos_ids_list.append(torch.arange(text_len).view(1, -1).expand(3, -1) + st_idx)

                llm_positions = torch.cat(llm_pos_ids_list, dim=1).reshape(3, -1)
                position_ids[..., i, attention_mask[i] == 1] = llm_positions.to(position_ids.device)
                mrope_position_deltas.append(llm_positions.max() + 1 - len(total_input_ids[i]))
            mrope_position_deltas = torch.tensor(mrope_position_deltas, device=input_ids.device).unsqueeze(1)
            return position_ids, mrope_position_deltas
        else:
            if attention_mask is not None:
                position_ids = attention_mask.long().cumsum(-1) - 1
                position_ids.masked_fill_(attention_mask == 0, 1)
                position_ids = position_ids.unsqueeze(0).expand(3, -1, -1).to(attention_mask.device)
                max_position_ids = position_ids.max(0, keepdim=False)[0].max(-1, keepdim=True)[0]
                mrope_position_deltas = max_position_ids + 1 - attention_mask.shape[-1]
            else:
                position_ids = (
                    torch.arange(input_ids.shape[1], device=input_ids.device)
                    .view(1, 1, -1)
                    .expand(3, input_ids.shape[0], -1)
                )
                mrope_position_deltas = torch.zeros(
                    [input_ids.shape[0], 1],
                    device=input_ids.device,
                    dtype=input_ids.dtype,
                )

            return position_ids, mrope_position_deltas

    def get_audio_features(
        self, audio_features: torch.FloatTensor
    ):
        audio_features = audio_features.type(self.audio.dtype)
        audio_embeds = self.audio(audio_features)
        return audio_embeds

    def get_video_features(
        self, pixel_values_videos: torch.FloatTensor, video_grid_thw: Optional[torch.LongTensor] = None
    ):
        """
        Encodes videos into continuous embeddings that can be forwarded to the language model. The deepstack visual features are also returned.

        Args:
            pixel_values_videos (`torch.FloatTensor` of shape `(batch_size, num_channels, image_size, image_size)`):
                The tensors corresponding to the input videos.
            video_grid_thw (`torch.LongTensor` of shape `(num_videos, 3)`, *optional*):
                The temporal, height and width of feature shape of each video in LLM.
        """
        # Same implementation as for images
        return self.get_image_features(pixel_values_videos, video_grid_thw)

    def get_image_features(self, pixel_values: torch.FloatTensor, image_grid_thw: Optional[torch.LongTensor] = None):
        """
        Encodes images into continuous embeddings that can be forwarded to the language model. The deepstack visual features are also returned.

        Args:
            pixel_values (`torch.FloatTensor` of shape `(batch_size, num_channels, image_size, image_size)`):
                The tensors corresponding to the input images.
            image_grid_thw (`torch.LongTensor` of shape `(num_images, 3)`, *optional*):
                The temporal, height and width of feature shape of each image in LLM.
        """
        pixel_values = pixel_values.type(self.visual.dtype)
        image_embeds, deepstack_image_embeds = self.visual(pixel_values, grid_thw=image_grid_thw)
        split_sizes = (image_grid_thw.prod(-1) // self.visual.spatial_merge_size**2).tolist()
        image_embeds = torch.split(image_embeds, split_sizes)
        return image_embeds, deepstack_image_embeds

    def get_placeholder_mask(
        self,
        input_ids: torch.LongTensor,
        inputs_embeds: torch.FloatTensor,
        image_features: Optional[torch.FloatTensor] = None,
        video_features: Optional[torch.FloatTensor] = None,
        audio_features: Optional[torch.FloatTensor] = None,
    ):
        """
        Obtains multimodal placeholder mask from `input_ids` or `inputs_embeds`, and checks that the placeholder token count is
        equal to the length of multimodal features. If the lengths are different, an error is raised.
        """
        if input_ids is None:
            special_image_mask = inputs_embeds == self.get_input_embeddings()(
                torch.tensor(self.config.image_token_id, dtype=torch.long, device=inputs_embeds.device)
            )
            special_image_mask = special_image_mask.all(-1)
            special_video_mask = inputs_embeds == self.get_input_embeddings()(
                torch.tensor(self.config.video_token_id, dtype=torch.long, device=inputs_embeds.device)
            )
            special_video_mask = special_video_mask.all(-1)
            special_audio_mask = inputs_embeds == self.get_input_embeddings()(
                torch.tensor(self.config.audio_token_id, dtype=torch.long, device=inputs_embeds.device)
            )
            special_audio_mask = special_audio_mask.all(-1)
        else:
            special_image_mask = input_ids == self.config.image_token_id
            special_video_mask = input_ids == self.config.video_token_id
            special_audio_mask = input_ids == self.config.audio_token_id

        n_image_tokens = special_image_mask.sum()
        special_image_mask = special_image_mask.unsqueeze(-1).expand_as(inputs_embeds).to(inputs_embeds.device)
        if image_features is not None and inputs_embeds[special_image_mask].numel() != image_features.numel():
            raise ValueError(
                f"Image features and image tokens do not match: tokens: {n_image_tokens}, features {image_features.shape[0]}"
            )

        n_video_tokens = special_video_mask.sum()
        special_video_mask = special_video_mask.unsqueeze(-1).expand_as(inputs_embeds).to(inputs_embeds.device)
        # if video_features is not None and inputs_embeds[special_video_mask].numel() != video_features.numel():
        # if (special_video_mask != 0).sum().item() != video_features.numel():
        #     raise ValueError(
        #         f"Videos features and video tokens do not match: tokens: {n_video_tokens}, features {video_features.shape[0]}"
        #     )

        n_audio_tokens = special_audio_mask.sum()
        special_audio_mask = special_audio_mask.unsqueeze(-1).expand_as(inputs_embeds).to(inputs_embeds.device)
        if audio_features is not None and inputs_embeds[special_audio_mask].numel() != audio_features.numel():
            raise ValueError(
                f"Audio features and audio tokens do not match: tokens: {n_audio_tokens}, features {audio_features.shape[0]}"
            )

        return special_image_mask, special_video_mask, special_audio_mask

    def recursive_attention(
        self,
        video_embeds,
        deepstack_video_embeds=None,
        downsample_ids=None,
        memory_size=None,
        stepsize=None,
        mergemode="",
    ):
        if deepstack_video_embeds is not None:
            embed_size = video_embeds.size(-1)
            video_embeds = torch.cat([video_embeds] + deepstack_video_embeds, dim=-1)
        if downsample_ids is None:
            downsample_ids = torch.tensor([n for n in range(0, video_embeds.size(0))]).to(video_embeds.device)
        memory_size = self.fixed_memory_size if memory_size is None else memory_size
        stepsize = self.stepsize if stepsize is None else stepsize
        running_embeds = video_embeds[:memory_size]
        running_downsample_ids = downsample_ids[:memory_size]
        for step_id in range(memory_size, video_embeds.size(0), stepsize):
            incoming_data = video_embeds[step_id:step_id+stepsize]
            incoming_downids = downsample_ids[step_id:step_id+stepsize]
            running_embeds = torch.cat([running_embeds, incoming_data], dim=0)
            fullsize = running_embeds.size(0)
            running_downsample_ids = torch.cat([running_downsample_ids, incoming_downids], dim=0)
            emb_norms = F.normalize(running_embeds, dim=-1)
            similarities = (emb_norms[:-1] * emb_norms[1:]).sum(dim=-1)
            top_sim_ids = (-similarities.squeeze(0)).topk(memory_size)[1] + 1
            top_sim_ids = torch.sort(top_sim_ids).values
            if "simsample" in self.ttt_type and mergemode != "sim":
                running_embeds = running_embeds[top_sim_ids]
            else:
                starts = torch.cat([top_sim_ids.new_zeros(1), top_sim_ids[:-1]], dim=0)
                lengths = top_sim_ids - starts
                row_idx = torch.arange(fullsize).unsqueeze(1).to(top_sim_ids.device)  # [N, 1]
                group_range = (row_idx >= starts) & (row_idx < top_sim_ids)   # [N, M]
                P_matrix = group_range.to(running_embeds.dtype) / lengths.to(running_embeds.dtype)
                running_embeds = torch.einsum("nm,nd->md", P_matrix, running_embeds)
            running_downsample_ids = running_downsample_ids[top_sim_ids.to(running_downsample_ids.device)]
        running_embeds_deepstack = None
        if deepstack_video_embeds is not None:
            running_embeds = running_embeds.view(running_embeds.size(0), -1, embed_size)
            running_embeds_deepstack = [running_embeds[:, i+1, :] for i in range(running_embeds.size(1) - 1)]
            running_embeds = running_embeds[:, 0, :]
        return running_embeds, running_embeds_deepstack, running_downsample_ids

    def ttt_padding(self, video_embeds):
        padding_length = 0
        minibatchsize = self.ttt_minibatch_size
        if video_embeds.size(1) % minibatchsize != 0:
            padding_length = minibatchsize - video_embeds.size(1) % minibatchsize
            padding = video_embeds.new_zeros(video_embeds.size(0), padding_length, video_embeds.size(-1))
            video_embeds = torch.cat([video_embeds, padding], dim=1)
        return video_embeds, padding_length

    def forward_ttt_layers(self, pixel_values_videos, video_grid_thw, input_embeds):
        if "ttt" in self.ttt_type:
            freqs_cis = self.ttt_layers._precompute_freqs_cis_3d(
                video_grid_thw[0, 1]//2, video_grid_thw[0, 2]//2, max(video_grid_thw[0, 0]*2, 128)).to(pixel_values_videos.device)

        if self.fixed_memory_size > 0:
            state_track = None
            running_downsample_ids, running_embeddings, running_deepstack_embeds = None, None, None
            stepsize = video_grid_thw[0, 1] * video_grid_thw[0, 2]
            step = 0
            num_frame_per_chunk = 256
            # with torch.no_grad():
            for i in range(0, video_grid_thw[0, 0], num_frame_per_chunk):
                current_thw = video_grid_thw.clone()
                current_thw[0, 0] = min(num_frame_per_chunk, video_grid_thw[0, 0] - i)
                video_embeds, incoming_deepstack_embeds = self.get_video_features(pixel_values_videos[i*stepsize:(i+num_frame_per_chunk)*stepsize], current_thw)
                video_embeds = torch.cat(video_embeds, dim=0).to(input_embeds.device, input_embeds.dtype)
                if "ttt" in self.ttt_type:
                    embed_dim = video_embeds.size(-1)
                    if "concat" in self.ttt_type:
                        video_embeds = torch.stack([video_embeds] + incoming_deepstack_embeds, dim=0)
                    else:
                        video_embeds = video_embeds.unsqueeze(0)
                    # residual_video_emb = video_embeds
                    incoming_embeddings, padding_length = self.ttt_padding(video_embeds)
                    incoming_embeddings, state_track = self.ttt_layers(
                        incoming_embeddings,
                        freqs_cis=freqs_cis[step:step+incoming_embeddings.size(1)] if freqs_cis is not None else None,
                        state_track=state_track,
                    )
                    if padding_length > 0:
                        incoming_embeddings = incoming_embeddings[:, :-padding_length]
                    if self.ttt_use_gating:
                        incoming_embeddings = self.ttt_gating(incoming_embeddings) + video_embeds
                    incoming_embeddings = incoming_embeddings.transpose(0, 1).reshape(incoming_embeddings.size(1), -1)
                else:
                    incoming_embeddings = video_embeds
                incoming_downsample_ids = torch.tensor([n for n in range(0, incoming_embeddings.size(0))]).to(video_embeds.device) + step
                step += incoming_downsample_ids.size(0)
                # print(step)

                if not self.training:
                    torch.cuda.empty_cache()

                if running_embeddings is not None:
                    running_embeddings = torch.cat([running_embeddings, incoming_embeddings], dim=0)
                    running_downsample_ids = torch.cat([running_downsample_ids, incoming_downsample_ids], dim=0)
                    if "concat" not in self.ttt_type:
                        running_deepstack_embeds = [torch.cat(
                            [r_emb, i_emb], dim=0) for r_emb, i_emb in zip(running_deepstack_embeds, incoming_deepstack_embeds)]
                        running_embeddings, running_deepstack_embeds, running_downsample_ids = self.recursive_attention(
                            running_embeddings, deepstack_video_embeds=running_deepstack_embeds, downsample_ids=running_downsample_ids)
                    else:
                        running_embeddings, _, running_downsample_ids = self.recursive_attention(
                            running_embeddings, downsample_ids=running_downsample_ids)
                else:
                    running_embeddings = incoming_embeddings
                    running_downsample_ids = incoming_downsample_ids
                    running_deepstack_embeds = incoming_deepstack_embeds
                    if incoming_embeddings.size(0) > self.fixed_memory_size:
                        if "concat" not in self.ttt_type:
                            running_embeddings, running_deepstack_embeds, running_downsample_ids = self.recursive_attention(
                                running_embeddings, deepstack_video_embeds=running_deepstack_embeds, downsample_ids=running_downsample_ids)
                        else:
                            running_embeddings, _, running_downsample_ids = self.recursive_attention(
                                running_embeddings, downsample_ids=running_downsample_ids)
            downsample_ids = running_downsample_ids
            video_embeds = running_embeddings
            deepstack_video_embeds = running_deepstack_embeds
        else:
            video_embeds, padding_length = self.ttt_padding(video_embeds)
            video_embeds, state_track = self.ttt_layers(video_embeds, freqs_cis=freqs_cis)

            if padding_length > 0:
                video_embeds = video_embeds[:, :-padding_length]

            if self.ttt_use_gating:
                video_embeds = self.ttt_gating(video_embeds) + residual_video_emb

            video_embeds, _, downsample_ids = self.recursive_attention(video_embeds.squeeze(0), downsample_ids=downsample_ids)
        # Split
        if "concat" in self.ttt_type:
            video_embeds = video_embeds.view(video_embeds.size(0), -1, embed_dim)
            deepstack_video_embeds = [video_embeds[:, i+1, :] for i in range(video_embeds.size(1) - 1)]
            video_embeds = video_embeds[:, 0, :]
        return video_embeds, deepstack_video_embeds, downsample_ids

    def compute_segment_logits_with_similarity_minmax(self, X, s, alpha=1.0, eps=1e-8):
        T, N = X.shape

        X_norm = F.normalize(X, p=2, dim=1, eps=eps)          # (T, N)

        # sim_forward[t] ~ cos(x_t, x_{t+1}) for t=0..T-2
        sim_forward = (X_norm[:-1] * X_norm[1:]).sum(dim=1)   # (T-1,)
        ones = torch.ones(1, device=X.device, dtype=X.dtype)
        sim_left  = torch.cat([ones, sim_forward])            # (T,)
        sim_right = torch.cat([sim_forward, ones])            # (T,)

        sim_mean = (sim_left + sim_right) / 2.0               # (T,)
        dissim = (1.0 - sim_mean) / 2.0                               # (T,) approx in [0, 2]

        # 2. Min-max normalize scores to [0, 1]
        s_min = s.min()
        s_max = s.max()
        s_norm = (s - s_min) / (s_max - s_min + eps)          # [0,1]

        simscaling = torch.sigmoid(self.search_alpha)

        z = (1 - simscaling) * s_norm + simscaling * dissim
        return z

    def gumbel_top_k(self, scores, K, tau=1.0, eps=1e-8):
        T = scores.shape[0]
        # 1) Sample Gumbel noise
        u = torch.rand_like(scores)
        gumbel = -torch.log(-torch.log(u + eps) + eps)

        # 2) Gumbel-perturbed scores
        perturbed = scores + gumbel

        # 3) Take Top-K
        topk_vals, topk_idx = torch.topk(perturbed, K, dim=0)  # (K,), (K,)

        # 4) Turn those K values into a softmax distribution (optional smoothing)
        topk_weights = F.softmax(topk_vals / tau, dim=0)       # (K,)

        # 5) Scatter back into a (T,) vector
        weights = torch.zeros_like(scores)
        weights[topk_idx] = topk_weights

        return topk_idx, weights

    def select_representatives_gumbel_top_k_ste(self, X, s, K, tau=1.0):
        # 1) Gumbel-Top-K selection
        idx, w = self.gumbel_top_k(s, K, tau=tau)   # idx: (K,), w: (T,)

        # 2) Hard representatives: actual selected rows
        reps_hard = X[idx]                     # (K, N)
        soft_vec = torch.matmul(w, X)          # (N,)
        reps_soft = soft_vec.unsqueeze(0).expand(K, -1)  # (K, N)

        reps = reps_soft + (reps_hard - reps_soft).detach()

        return reps, idx, w

    def importance_pool(self, video_embed, importance, memory_size):
        T, N = video_embed.shape
        if "gumbel" not in self.search_type:
            if "softmax" in self.search_type:
                importance = torch.softmax(importance, dim=-1)
            elif "sigmoid" in self.search_type:
                importance = torch.sigmoid(importance)
            importance = self.compute_segment_logits_with_similarity_minmax(video_embed, importance)
            top_sim_ids = torch.sort(importance.topk(memory_size)[1]).values
            video_embed = importance.unsqueeze(1) * video_embed
            starts = torch.cat([top_sim_ids.new_zeros(1) - 1, top_sim_ids[:-1]], dim=0)
            lengths = top_sim_ids - starts
            row_idx = torch.arange(T).unsqueeze(1).to(top_sim_ids.device)  # [N, 1]
            group_range = (row_idx > starts) & (row_idx <= top_sim_ids)   # [N, M]
            P_matrix = group_range.to(video_embed.dtype) / lengths.to(video_embed.dtype)
            video_embed = torch.einsum("nm,nd->md", P_matrix, video_embed)
        else:
            importance = self.compute_segment_logits_with_similarity_minmax(video_embed, importance)
            video_embed, top_sim_ids, weights = self.select_representatives_gumbel_top_k_ste(video_embed, importance, memory_size)
        return video_embed, top_sim_ids

    def chunk_memory(self, video_embeds, deepstack_video_embeds, downsample_ids, hidden_states):
        if "attn" in self.search_type:
            attention_scores = torch.einsum("k,ijk->ij", self.search_query, hidden_states)
            attention_scores = F.softmax(attention_scores / math.sqrt(self.search_query.size(0)), dim=-1)
            search_query = torch.einsum("ij,ijk->ik", attention_scores, hidden_states).unsqueeze(1)
        else:
            search_query = hidden_states.sum(dim=1, keepdim=True)
        scores = torch.einsum("ij,j->i", video_embeds, search_query.squeeze(0).squeeze(0)) / math.sqrt(self.hidden_size)
        memsize = min(video_embeds.size(0), self.workingmemsize)
        video_embeds, local_downsample_ids = self.importance_pool(video_embeds, scores, memsize)
        deepstack_video_embeds = [emb[local_downsample_ids] for emb in deepstack_video_embeds]
        downsample_ids = downsample_ids[local_downsample_ids]
        return video_embeds, deepstack_video_embeds, downsample_ids

    @auto_docstring
    @check_model_inputs
    def forward(
        self,
        input_ids: torch.LongTensor = None,
        attention_mask: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.LongTensor] = None,
        past_key_values: Optional[Cache] = None,
        inputs_embeds: Optional[torch.FloatTensor] = None,
        pixel_values: Optional[torch.Tensor] = None,
        pixel_values_videos: Optional[torch.FloatTensor] = None,
        image_grid_thw: Optional[torch.LongTensor] = None,
        video_grid_thw: Optional[torch.LongTensor] = None,
        cache_position: Optional[torch.LongTensor] = None,
        audio_feature: Optional[torch.Tensor] = None,
        distillround: bool = False,
        labels: Optional[torch.LongTensor] = None,
        memory_triplets: Optional[torch.Tensor] = None,
        **kwargs: Unpack[TransformersKwargs],
    ) -> Union[tuple, Qwen3VLModelOutputWithPast]:
        r"""
        image_grid_thw (`torch.LongTensor` of shape `(num_images, 3)`, *optional*):
            The temporal, height and width of feature shape of each image in LLM.
        video_grid_thw (`torch.LongTensor` of shape `(num_videos, 3)`, *optional*):
            The temporal, height and width of feature shape of each video in LLM.
        """
        if (input_ids is None) ^ (inputs_embeds is not None):
            raise ValueError("You must specify exactly one of input_ids or inputs_embeds")

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

        image_mask = None
        video_mask = None

        if pixel_values is not None:
            image_embeds, deepstack_image_embeds = self.get_image_features(pixel_values, image_grid_thw)
            image_embeds = torch.cat(image_embeds, dim=0).to(inputs_embeds.device, inputs_embeds.dtype)
            image_mask, _ = self.get_placeholder_mask(
                input_ids, inputs_embeds=inputs_embeds, image_features=image_embeds
            )
            inputs_embeds = inputs_embeds.masked_scatter(image_mask, image_embeds)

        all_memory = None
        if pixel_values_videos is not None:
            if self.fixed_memory_size > 0:
                if memory_triplets is not None:
                    video_embeds, deepstack_video_embeds, downsample_ids = memory_triplets["video_embeds"], memory_triplets["deepstack_video_embeds"], memory_triplets["downsample_ids"]
                elif self.freeze_ttt:
                    with torch.no_grad():
                        video_embeds, deepstack_video_embeds, downsample_ids = self.forward_ttt_layers(pixel_values_videos, video_grid_thw, inputs_embeds)
                        if self.search_type != "none" and memory_triplets is None:
                            all_memory = {
                                "video_embeds": video_embeds,
                                "deepstack_video_embeds": deepstack_video_embeds,
                                "downsample_ids": downsample_ids,
                            }
                            memsize = self.workingmemsize // 8
                            video_embeds, deepstack_video_embeds, downsample_ids = self.recursive_attention(
                                video_embeds, deepstack_video_embeds=deepstack_video_embeds, downsample_ids=downsample_ids, memory_size=memsize)
                else:
                    video_embeds, deepstack_video_embeds, downsample_ids = self.forward_ttt_layers(pixel_values_videos, video_grid_thw, inputs_embeds)
                    if self.search_type != "none" and memory_triplets is None:
                        all_memory = {
                            "video_embeds": video_embeds,
                            "deepstack_video_embeds": deepstack_video_embeds,
                            "downsample_ids": downsample_ids,
                        }
                        memsize = self.workingmemsize // 8
                        video_embeds, deepstack_video_embeds, downsample_ids = self.recursive_attention(
                            video_embeds, deepstack_video_embeds=deepstack_video_embeds, downsample_ids=downsample_ids, memory_size=memsize)
                _, video_mask, _ = self.get_placeholder_mask(
                    input_ids, inputs_embeds=inputs_embeds, video_features=video_embeds
                )
            else:
                video_embeds, deepstack_video_embeds = self.get_video_features(pixel_values_videos, video_grid_thw)
                video_embeds = torch.cat(video_embeds, dim=0).to(inputs_embeds.device, inputs_embeds.dtype)
                _, video_mask, _ = self.get_placeholder_mask(
                    input_ids, inputs_embeds=inputs_embeds, video_features=video_embeds
                )
                inputs_embeds = inputs_embeds.masked_scatter(video_mask, video_embeds)
            if torch.cuda.current_device() == 0:
                print(f"RANK 0 video embeds shape: {video_embeds.shape}")

        audio_downsample_ids = None
        if audio_feature is not None:
            audio_embeds = self.get_audio_features(audio_feature)
            if torch.cuda.current_device() == 0:
                print(f"RANK 0 audio embeds shape: {audio_embeds.shape}")
            _, _, audio_mask = self.get_placeholder_mask(
                input_ids, inputs_embeds=inputs_embeds, audio_features=audio_embeds
            )
            if self.fixed_memory_size > 0 and not distillround:
                audio_embeds, _, audio_downsample_ids = self.recursive_attention(
                    audio_embeds.view(-1, audio_embeds.size(-1)),
                    memory_size=self.fixed_memory_size_audio,
                    stepsize=self.stepsize,
                    mergemode="sim",
                )
            else:
                inputs_embeds = inputs_embeds.masked_scatter(audio_mask, audio_embeds)
        # print(f"Both shapes: {video_embeds.shape}, {audio_embeds.shape}")

        visual_pos_masks = None
        deepstack_visual_embeds = None
        if image_mask is not None and video_mask is not None:
            # aggregate visual_pos_masks and deepstack_visual_embeds
            image_mask = image_mask[..., 0]
            video_mask = video_mask[..., 0]
            visual_pos_masks = image_mask | video_mask
            deepstack_visual_embeds = []
            image_mask_joint = image_mask[visual_pos_masks]
            video_mask_joint = video_mask[visual_pos_masks]
            for img_embed, vid_embed in zip(deepstack_image_embeds, deepstack_video_embeds):
                embed_joint = img_embed.new_zeros(visual_pos_masks.sum(), img_embed.shape[-1]).to(img_embed.device)
                embed_joint[image_mask_joint, :] = img_embed
                embed_joint[video_mask_joint, :] = vid_embed
                deepstack_visual_embeds.append(embed_joint)
        elif image_mask is not None:
            image_mask = image_mask[..., 0]
            visual_pos_masks = image_mask
            deepstack_visual_embeds = deepstack_image_embeds
        elif video_mask is not None:
            visual_pos_masks = video_mask[..., 0]
            deepstack_visual_embeds = deepstack_video_embeds

        if position_ids is None:
            attention_mask_tensor = (
                attention_mask if not isinstance(attention_mask, dict) else attention_mask["full_attention"]
            )
            if attention_mask_tensor is not None and attention_mask_tensor.ndim == 4:
                attention_mask_tensor = torch.diagonal(attention_mask_tensor[:, 0], dim1=1, dim2=2)
                # Only apply conversion for floating point tensors (inverted masks)
                if attention_mask_tensor.dtype.is_floating_point:
                    attention_mask_tensor = attention_mask_tensor / torch.finfo(attention_mask_tensor.dtype).min
                    attention_mask_tensor = (1.0 - attention_mask_tensor).int()

            # Calculate RoPE index once per generation in the pre-fill stage only.
            # When compiling, we can't check tensor values thus we check only input length
            # It is safe to assume that `length!=1` means we're in pre-fill because compiled
            # models currently cannot do asssisted decoding
            prefill_compiled_stage = is_torchdynamo_compiling() and (
                (input_ids is not None and input_ids.shape[1] != 1)
                or (inputs_embeds is not None and inputs_embeds.shape[1] != 1)
            )
            prefill_noncompiled_stage = not is_torchdynamo_compiling() and (
                (cache_position is not None and cache_position[0] == 0)
                or (past_key_values is None or past_key_values.get_seq_length() == 0)
            )
            if (prefill_compiled_stage or prefill_noncompiled_stage) or self.rope_deltas is None:
                position_ids, rope_deltas = self.get_rope_index(
                    input_ids,
                    image_grid_thw,
                    video_grid_thw,
                    attention_mask=attention_mask_tensor,
                )
                self.rope_deltas = rope_deltas
            # then use the prev pre-calculated rope-deltas to get the correct position ids
            else:
                batch_size, seq_length, _ = inputs_embeds.shape
                delta = (
                    (cache_position[0] + self.rope_deltas).to(inputs_embeds.device)
                    if cache_position is not None
                    else 0
                )
                position_ids = torch.arange(seq_length, device=inputs_embeds.device)
                position_ids = position_ids.view(1, -1).expand(batch_size, -1)
                if cache_position is not None:  # otherwise `deltas` is an int `0`
                    delta = delta.repeat_interleave(batch_size // delta.shape[0], dim=0)
                position_ids = position_ids.add(delta)
                position_ids = position_ids.unsqueeze(0).expand(3, -1, -1)

        if self.fixed_memory_size > 0 and pixel_values_videos is not None and not distillround:
            video_mask_sel = torch.where(visual_pos_masks == 1)[1][downsample_ids]
            if audio_feature is not None and audio_downsample_ids is not None:
                audio_mask_sel = torch.where(audio_mask[:, :, 0] == 1)[1][audio_downsample_ids]
                keep_mask = ~(audio_mask[:, :, 0] + visual_pos_masks)
                keep_mask[:, video_mask_sel] = True
                keep_mask[:, audio_mask_sel] = True
                keep_ids = torch.where(keep_mask)[1]
                inputs_embeds = inputs_embeds[:, keep_ids]
                inputs_embeds = inputs_embeds.masked_scatter(video_mask[:, keep_ids], video_embeds)
                inputs_embeds = inputs_embeds.masked_scatter(audio_mask[:, keep_ids], audio_embeds)
            else:
                keep_mask = ~visual_pos_masks
                keep_mask[:, video_mask_sel] = True
                keep_ids = torch.where(keep_mask)[1]
                inputs_embeds = inputs_embeds[:, keep_ids]
                inputs_embeds = inputs_embeds.masked_scatter(video_mask[:, keep_ids], video_embeds)
            if labels is not None:
                labels = labels[:, keep_ids]
            attention_mask = attention_mask[:, keep_ids]
            visual_pos_masks = visual_pos_masks[:, keep_ids]
            newposids = []
            for posids in position_ids:
                newposid = posids[:, keep_ids]
                newposids.append(newposid)
            position_ids = torch.stack(newposids, dim=0)

        outputs = self.language_model(
            input_ids=None,
            position_ids=position_ids,
            attention_mask=attention_mask,
            past_key_values=past_key_values,
            inputs_embeds=inputs_embeds,
            cache_position=cache_position,
            visual_pos_masks=visual_pos_masks,
            deepstack_visual_embeds=deepstack_visual_embeds,
            **kwargs,
        )
        if self.search_type != "none" and all_memory is not None:
            query_start_pos = torch.where(input_ids == 151653)[1][-1].item() + 2 - input_ids.size(1)
            if len(torch.where(input_ids == 151644)[1]) > 2:
                gen_start_pos = torch.where(input_ids == 151644)[1][2].item() + 2 - input_ids.size(1)
            else:
                gen_start_pos = -1
            hidden_states = outputs.last_hidden_state[:, query_start_pos:gen_start_pos]
            video_embeds, deepstack_video_embeds, downsample_ids = self.chunk_memory(
                all_memory["video_embeds"],
                all_memory["deepstack_video_embeds"],
                all_memory["downsample_ids"],
                hidden_states,
            )
            memory_triplets = {
                "video_embeds": video_embeds,
                "deepstack_video_embeds": deepstack_video_embeds,
                "downsample_ids": downsample_ids,
            }
        else:
            memory_triplets = None
            torch.cuda.empty_cache()

        return Qwen3VLModelOutputWithPast(
            last_hidden_state=outputs.last_hidden_state,
            past_key_values=outputs.past_key_values,
            rope_deltas=self.rope_deltas,
            labels=labels,
            memory_triplets=memory_triplets,
        )


@dataclass
@auto_docstring(
    custom_intro="""
    Base class for Qwen3VL causal language model (or autoregressive) outputs.
    """
)
class Qwen3VLCausalLMOutputWithPast(ModelOutput):
    r"""
    loss (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `labels` is provided):
        Language modeling loss (for next-token prediction).
    logits (`torch.FloatTensor` of shape `(batch_size, sequence_length, config.vocab_size)`):
        Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).
    past_key_values (`Cache`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
        It is a [`~cache_utils.Cache`] instance. For more details, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache).

        Contains pre-computed hidden-states (key and values in the self-attention blocks) that can be used (see
        `past_key_values` input) to speed up sequential decoding.
    rope_deltas (`torch.LongTensor` of shape `(batch_size, )`, *optional*):
        The rope index difference between sequence length and multimodal rope.
    """

    loss: Optional[torch.FloatTensor] = None
    logits: Optional[torch.FloatTensor] = None
    past_key_values: Optional[Cache] = None
    hidden_states: Optional[tuple[torch.FloatTensor]] = None
    attentions: Optional[tuple[torch.FloatTensor]] = None
    rope_deltas: Optional[torch.LongTensor] = None
    memory_triplets: Optional[Dict[str, torch.Tensor]] = None


class Qwen3VLForConditionalGeneration(Qwen3VLPreTrainedModel, GenerationMixin):
    _checkpoint_conversion_mapping = {}
    _tied_weights_keys = ["lm_head.weight"]
    # Reference: fix gemma3 grad acc #37208
    accepts_loss_kwargs = False
    config: Qwen3VLConfig

    def __init__(self, config):
        super().__init__(config)
        self.model = Qwen3VLModel(config)
        self.lm_head = nn.Linear(config.text_config.hidden_size, config.text_config.vocab_size, bias=False)

        self.post_init()

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

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

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

    def get_decoder(self):
        return self.model.get_decoder()

    def get_video_features(
        self, pixel_values_videos: torch.FloatTensor, video_grid_thw: Optional[torch.LongTensor] = None
    ):
        return self.model.get_video_features(pixel_values_videos, video_grid_thw)

    def get_image_features(self, pixel_values: torch.FloatTensor, image_grid_thw: Optional[torch.LongTensor] = None):
        return self.model.get_image_features(pixel_values, image_grid_thw)

    # Make modules available through conditional class for BC
    @property
    def language_model(self):
        return self.model.language_model

    @property
    def visual(self):
        return self.model.visual

    @check_model_inputs
    def forward(
        self,
        input_ids: torch.LongTensor = None,
        attention_mask: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.LongTensor] = None,
        past_key_values: Optional[Cache] = None,
        inputs_embeds: Optional[torch.FloatTensor] = None,
        labels: Optional[torch.LongTensor] = None,
        pixel_values: Optional[torch.Tensor] = None,
        pixel_values_videos: Optional[torch.FloatTensor] = None,
        image_grid_thw: Optional[torch.LongTensor] = None,
        video_grid_thw: Optional[torch.LongTensor] = None,
        cache_position: Optional[torch.LongTensor] = None,
        logits_to_keep: Union[int, torch.Tensor] = 0,
        audio_feature: Optional[torch.Tensor] = None,
        train_type: Optional[str] = "sft",
        distillround: bool = False,
        memory_triplets: Optional[dict] = None,
        **kwargs: Unpack[TransformersKwargs],
    ) -> Union[tuple, Qwen3VLCausalLMOutputWithPast]:
        r"""
        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]`.
        image_grid_thw (`torch.LongTensor` of shape `(num_images, 3)`, *optional*):
            The temporal, height and width of feature shape of each image in LLM.
        video_grid_thw (`torch.LongTensor` of shape `(num_videos, 3)`, *optional*):
            The temporal, height and width of feature shape of each video in LLM.

        Example:
            TODO: Add example
        """
        outputs = self.model(
            input_ids=input_ids,
            pixel_values=pixel_values,
            pixel_values_videos=pixel_values_videos,
            image_grid_thw=image_grid_thw,
            video_grid_thw=video_grid_thw,
            position_ids=position_ids,
            attention_mask=attention_mask,
            past_key_values=past_key_values,
            inputs_embeds=inputs_embeds,
            cache_position=cache_position,
            audio_feature=audio_feature,
            labels=labels,
            distillround=distillround,
            memory_triplets=memory_triplets,
            **kwargs,
        )

        if outputs.labels is not None:
            labels = outputs.labels

        hidden_states = outputs[0]

        loss = None
        logits = None

        shift_labels = kwargs.pop("shift_labels", None)
        return_logits = kwargs.pop("return_logits", False)
        memory_triplets = outputs.memory_triplets

        if self.training and (labels is not None or shift_labels is not None):
            loss = LigerForCausalLMLoss(
                hidden_states=hidden_states,
                lm_head_weight=self.lm_head.weight,
                labels=labels,
                shift_labels=shift_labels,
                hidden_size=self.config.text_config.hidden_size,
                **kwargs,
            )
            if return_logits:
                distill_labels = labels[0, 1:]
                start_idx = torch.where((distill_labels != -100)==True)[0][0]
                logits = self.lm_head(hidden_states[:, start_idx:, :])
        elif memory_triplets is None:
            # Only compute necessary logits, and do not upcast them to float if we are not computing the loss
            slice_indices = slice(-logits_to_keep, None) if isinstance(logits_to_keep, int) else logits_to_keep
            logits = self.lm_head(hidden_states[:, slice_indices, :])

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

        return Qwen3VLCausalLMOutputWithPast(
            loss=loss,
            logits=logits,
            past_key_values=outputs.past_key_values,
            rope_deltas=outputs.rope_deltas,
            memory_triplets=memory_triplets,
        )

    def prepare_inputs_for_generation(
        self,
        input_ids,
        past_key_values=None,
        attention_mask=None,
        inputs_embeds=None,
        cache_position=None,
        position_ids=None,
        use_cache=True,
        pixel_values=None,
        pixel_values_videos=None,
        image_grid_thw=None,
        video_grid_thw=None,
        audio_feature=None,
        **kwargs,
    ):
        # Overwritten -- in specific circumstances we don't want to forward image inputs to the model

        model_inputs = super().prepare_inputs_for_generation(
            input_ids,
            past_key_values=past_key_values,
            attention_mask=attention_mask,
            inputs_embeds=inputs_embeds,
            cache_position=cache_position,
            position_ids=position_ids,
            pixel_values=pixel_values,
            pixel_values_videos=pixel_values_videos,
            image_grid_thw=image_grid_thw,
            video_grid_thw=video_grid_thw,
            use_cache=use_cache,
            audio_feature=audio_feature,
            **kwargs,
        )

        # Qwen3VL position_ids are prepareed with rope_deltas in forward
        model_inputs["position_ids"] = None

        if cache_position[0] != 0:
            model_inputs["pixel_values"] = None
            model_inputs["pixel_values_videos"] = None
            model_inputs["audio_feature"] = None

        return model_inputs

    def _get_image_nums_and_video_nums(
        self,
        input_ids: Optional[torch.LongTensor],
        inputs_embeds: Optional[torch.Tensor] = None,
    ) -> tuple[torch.Tensor, torch.Tensor]:
        """
        Get the number of images and videos for each sample to calculate the separation length of the sample tensor.
        These parameters are not passed through the processor to avoid unpredictable impacts from interface modifications.

        Args:
            input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
                Indices of input sequence tokens in the vocabulary.

        Returns:
            image_nums (`torch.LongTensor` of shape `(batch_size, num_images_sample)`)
            video_nums (`torch.LongTensor` of shape `(batch_size, num_videos_sample)`)
        """
        image_token_id = self.config.image_token_id
        video_token_id = self.config.video_token_id
        vision_start_token_id = self.config.vision_start_token_id

        if inputs_embeds is not None:
            vision_start_mask = (
                inputs_embeds
                == self.get_input_embeddings()(
                    torch.tensor(vision_start_token_id, dtype=torch.long, device=inputs_embeds.device)
                )
            )[..., 0]
            image_mask = (
                inputs_embeds
                == self.get_input_embeddings()(
                    torch.tensor(image_token_id, dtype=torch.long, device=inputs_embeds.device)
                )
            )[..., 0]
            video_mask = (
                inputs_embeds
                == self.get_input_embeddings()(
                    torch.tensor(video_token_id, dtype=torch.long, device=inputs_embeds.device)
                )
            )[..., 0]
        else:
            vision_start_mask = input_ids == vision_start_token_id
            image_mask = input_ids == image_token_id
            video_mask = input_ids == video_token_id

        vision_first_mask = torch.roll(vision_start_mask, shifts=1, dims=1)
        image_nums = torch.sum(vision_first_mask & image_mask, dim=1)
        video_nums = torch.sum(vision_first_mask & video_mask, dim=1)

        return image_nums, video_nums

    def _expand_inputs_for_generation(
        self,
        expand_size: int = 1,
        is_encoder_decoder: bool = False,
        input_ids: Optional[torch.LongTensor] = None,
        **model_kwargs,
    ) -> tuple[torch.LongTensor, dict[str, Any]]:
        # Overwritten -- Support for expanding tensors without a batch size dimension
        # e.g., pixel_values, image_grid_thw, pixel_values_videos, video_grid_thw, second_per_grid_t
        # pixel_values.shape[0] is sum(seqlen_images for samples)
        # image_grid_thw.shape[0] is sum(num_images for samples)

        if expand_size == 1:
            return input_ids, model_kwargs

        visual_keys = ["pixel_values", "image_grid_thw", "pixel_values_videos", "video_grid_thw", "second_per_grid_ts"]

        def _expand_dict_for_generation_visual(dict_to_expand):
            image_grid_thw = model_kwargs.get("image_grid_thw", None)
            video_grid_thw = model_kwargs.get("video_grid_thw", None)
            image_nums, video_nums = self._get_image_nums_and_video_nums(
                input_ids, inputs_embeds=model_kwargs.get("inputs_embeds", None)
            )

            def _repeat_interleave_samples(x, lengths, repeat_times):
                samples = torch.split(x, lengths)
                repeat_args = [repeat_times] + [1] * (x.dim() - 1)
                result = torch.cat([sample.repeat(*repeat_args) for sample in samples], dim=0)
                return result

            for key in dict_to_expand:
                if key == "pixel_values":
                    # split images into samples
                    samples = torch.split(image_grid_thw, list(image_nums))
                    # compute the sequence length of images for each sample
                    lengths = [torch.prod(sample, dim=1).sum() for sample in samples]
                    dict_to_expand[key] = _repeat_interleave_samples(
                        dict_to_expand[key], lengths=lengths, repeat_times=expand_size
                    )
                elif key == "image_grid_thw":
                    # get the num of images for each sample
                    lengths = list(image_nums)
                    dict_to_expand[key] = _repeat_interleave_samples(
                        dict_to_expand[key], lengths=lengths, repeat_times=expand_size
                    )
                elif key == "pixel_values_videos":
                    samples = torch.split(video_grid_thw, list(video_nums))
                    lengths = [torch.prod(sample, dim=1).sum() for sample in samples]
                    dict_to_expand[key] = _repeat_interleave_samples(
                        dict_to_expand[key], lengths=lengths, repeat_times=expand_size
                    )
                elif key == "video_grid_thw":
                    lengths = list(video_nums)
                    dict_to_expand[key] = _repeat_interleave_samples(
                        dict_to_expand[key], lengths=lengths, repeat_times=expand_size
                    )
                elif key == "second_per_grid_ts":
                    dict_to_expand[key] = _repeat_interleave_samples(
                        dict_to_expand[key], lengths=list(video_nums), repeat_times=expand_size
                    )
            return dict_to_expand

        def _expand_dict_for_generation(dict_to_expand):
            for key in dict_to_expand:
                if (
                    key != "cache_position"
                    and dict_to_expand[key] is not None
                    and isinstance(dict_to_expand[key], torch.Tensor)
                    and key not in visual_keys
                ):
                    dict_to_expand[key] = dict_to_expand[key].repeat_interleave(expand_size, dim=0)
            return dict_to_expand

        model_kwargs = _expand_dict_for_generation_visual(model_kwargs)

        if input_ids is not None:
            input_ids = input_ids.repeat_interleave(expand_size, dim=0)

        model_kwargs = _expand_dict_for_generation(model_kwargs)

        if is_encoder_decoder:
            if model_kwargs.get("encoder_outputs") is None:
                raise ValueError("If `is_encoder_decoder` is True, make sure that `encoder_outputs` is defined.")
            model_kwargs["encoder_outputs"] = _expand_dict_for_generation(model_kwargs["encoder_outputs"])

        return input_ids, model_kwargs


__all__ = [
    "Qwen3VLVisionModel",
    "Qwen3VLForConditionalGeneration",
    "Qwen3VLModel",
    "Qwen3VLPreTrainedModel",
    "Qwen3VLTextModel",
]