video_diffusion/models/attention.py

# code mostly taken from https://github.com/huggingface/diffusers
from dataclasses import dataclass
from typing import Optional, Callable
import math
import torch
from torch import nn
import torch.nn.functional as F
from diffusers.configuration_utils import ConfigMixin, register_to_config
from diffusers import ModelMixin
from diffusers.models.cross_attention import CrossAttention
from diffusers.models.attention import FeedForward, AdaLayerNorm
from diffusers.utils import BaseOutput
from diffusers.utils.import_utils import is_xformers_available

from einops import rearrange
import numpy as np
import os 
from PIL import Image
import glob


@dataclass
class SpatioTemporalTransformerModelOutput(BaseOutput):
    """torch.FloatTensor of shape [batch x channel x frames x height x width]"""

    sample: torch.FloatTensor


if is_xformers_available():
    import xformers
    import xformers.ops
else:
    xformers = None


class SpatioTemporalTransformerModel(ModelMixin, ConfigMixin):
    @register_to_config
    def __init__(
        self,
        num_attention_heads: int = 16,
        attention_head_dim: int = 88,
        in_channels: Optional[int] = None,
        num_layers: int = 1,
        dropout: float = 0.0,
        norm_num_groups: int = 32,
        cross_attention_dim: Optional[int] = None,
        attention_bias: bool = False,
        activation_fn: str = "geglu",
        num_embeds_ada_norm: Optional[int] = None,
        use_linear_projection: bool = False,
        only_cross_attention: bool = False,
        upcast_attention: bool = False,
        use_temporal: bool = True,
        model_config: dict = {},
        **transformer_kwargs,
    ):
        super().__init__()
        self.use_linear_projection = use_linear_projection
        self.num_attention_heads = num_attention_heads
        self.attention_head_dim = attention_head_dim
        inner_dim = num_attention_heads * attention_head_dim

        # Define input layers
        self.in_channels = in_channels

        self.norm = torch.nn.GroupNorm(
            num_groups=norm_num_groups, num_channels=in_channels, eps=1e-6, affine=True
        )
        if use_linear_projection:
            self.proj_in = nn.Linear(in_channels, inner_dim)
        else:
            self.proj_in = nn.Conv2d(in_channels, inner_dim, kernel_size=1, stride=1, padding=0)

        # Define transformers blocks
        self.transformer_blocks = nn.ModuleList(
            [
                SpatioTemporalTransformerBlock(
                    inner_dim,
                    num_attention_heads,
                    attention_head_dim,
                    dropout=dropout,
                    cross_attention_dim=cross_attention_dim,
                    activation_fn=activation_fn,
                    num_embeds_ada_norm=num_embeds_ada_norm,
                    attention_bias=attention_bias,
                    only_cross_attention=only_cross_attention,
                    upcast_attention=upcast_attention,
                    use_temporal=use_temporal,
                    model_config=model_config,
                    **transformer_kwargs,
                )
                for d in range(num_layers)
            ]
        )

        # Define output layers
        if use_linear_projection:
            self.proj_out = nn.Linear(in_channels, inner_dim)
        else:
            self.proj_out = nn.Conv2d(inner_dim, in_channels, kernel_size=1, stride=1, padding=0)

    def forward(
        self, hidden_states, encoder_hidden_states=None, timestep=None, return_dict: bool = True, **kwargs,
    ):
        # 1. Input
        clip_length = None
        is_video = hidden_states.ndim == 5
        if is_video:
            clip_length = hidden_states.shape[2]
            hidden_states = rearrange(hidden_states, "b c f h w -> (b f) c h w")
            encoder_hidden_states = encoder_hidden_states.repeat_interleave(clip_length, 0)
        else:
            # To adapt to classifier-free guidance where encoder_hidden_states=2
            batch_size = hidden_states.shape[0]//encoder_hidden_states.shape[0]
            encoder_hidden_states = encoder_hidden_states.repeat_interleave(batch_size, 0)
        *_, h, w = hidden_states.shape
        residual = hidden_states

        # check resolution
 
        height = hidden_states.shape[-2]
        width = hidden_states.shape[-1]
        trajs = {}
        trajs["traj"] = kwargs["trajs"]["traj{}".format(height)]
        trajs["mask"] = kwargs["trajs"]["mask{}".format(height)]
        # trajs["t"] = kwargs["t"]
        trajs["old_qk"] = kwargs["old_qk"]
        trajs['flatten_res'] = kwargs['flatten_res']
        trajs['height'] = height
        trajs['width'] = width

        hidden_states = self.norm(hidden_states)
        if not self.use_linear_projection:
            hidden_states = self.proj_in(hidden_states)
            hidden_states = rearrange(hidden_states, "b c h w -> b (h w) c") # (bf) (hw) c
        else:
            hidden_states = rearrange(hidden_states, "b c h w -> b (h w) c")
            hidden_states = self.proj_in(hidden_states)

        # 2. Blocks
        for block in self.transformer_blocks:
            hidden_states = block(
                hidden_states, # [16, 4096, 320]
                encoder_hidden_states=encoder_hidden_states, # ([1, 77, 768]
                timestep=timestep,
                clip_length=clip_length,
                **trajs,
            )

        # 3. Output
        if not self.use_linear_projection:
            hidden_states = rearrange(hidden_states, "b (h w) c -> b c h w", h=h, w=w).contiguous()
            hidden_states = self.proj_out(hidden_states)
        else:
            hidden_states = self.proj_out(hidden_states)
            hidden_states = rearrange(hidden_states, "b (h w) c -> b c h w", h=h, w=w).contiguous()

        output = hidden_states + residual
        if is_video:
            output = rearrange(output, "(b f) c h w -> b c f h w", f=clip_length)

        if not return_dict:
            return (output,)

        return SpatioTemporalTransformerModelOutput(sample=output)

import copy
class SpatioTemporalTransformerBlock(nn.Module):
    def __init__(
        self,
        dim: int,
        num_attention_heads: int,
        attention_head_dim: int,
        dropout=0.0,
        cross_attention_dim: Optional[int] = None,
        activation_fn: str = "geglu",
        num_embeds_ada_norm: Optional[int] = None,
        attention_bias: bool = False,
        only_cross_attention: bool = False,
        upcast_attention: bool = False,
        use_sparse_causal_attention: bool = True,
        use_temporal:bool = False,
        temporal_attention_position: str = "after_feedforward",
        model_config: dict = {}
    ):
        super().__init__()
        self.use_temporal=use_temporal
        self.only_cross_attention = only_cross_attention
        self.use_ada_layer_norm = num_embeds_ada_norm is not None
        self.use_sparse_causal_attention = use_sparse_causal_attention
        # For safety, freeze the model_config
        self.model_config = copy.deepcopy(model_config)
        if 'least_sc_channel' in model_config:
            if dim< model_config['least_sc_channel']:
                self.model_config['SparseCausalAttention_index'] = []
        
        self.temporal_attention_position = temporal_attention_position
        temporal_attention_positions = ["after_spatial", "after_cross", "after_feedforward"]
        if temporal_attention_position not in temporal_attention_positions:
            raise ValueError(
                f"`temporal_attention_position` must be one of {temporal_attention_positions}"
            )

        # 1. Spatial-Attn
        # spatial_attention = SparseCausalAttention if use_sparse_causal_attention else CrossAttention
        # self.attn1 = spatial_attention(
        #     query_dim=dim,
        #     heads=num_attention_heads,
        #     dim_head=attention_head_dim,
        #     dropout=dropout,
        #     bias=attention_bias,
        #     cross_attention_dim=cross_attention_dim if only_cross_attention else None,
        #     upcast_attention=upcast_attention,
        # )  
        # is a self-attention
            # Fully
        self.attn1 = FullyFrameAttention(
            query_dim=dim,
            heads=num_attention_heads,
            dim_head=attention_head_dim,
            dropout=dropout,
            bias=attention_bias,
            cross_attention_dim=cross_attention_dim if only_cross_attention else None,
            upcast_attention=upcast_attention,
        )
        self.norm1 = (
            AdaLayerNorm(dim, num_embeds_ada_norm) if self.use_ada_layer_norm else nn.LayerNorm(dim)
        )

        # 2. Cross-Attn
        if cross_attention_dim is not None:
            self.attn2 = CrossAttention(
                query_dim=dim,
                cross_attention_dim=cross_attention_dim,
                heads=num_attention_heads,
                dim_head=attention_head_dim,
                dropout=dropout,
                bias=attention_bias,
                upcast_attention=upcast_attention,
            )  # is self-attn if encoder_hidden_states is none
            self.norm2 = (
                AdaLayerNorm(dim, num_embeds_ada_norm) if self.use_ada_layer_norm else nn.LayerNorm(dim)
            )
        else:
            self.attn2 = None
            self.norm2 = None

        # 3. Temporal-Attn
        if use_temporal:
            self.attn_temporal = CrossAttention(
                query_dim=dim,
                heads=num_attention_heads,
                dim_head=attention_head_dim,
                dropout=dropout,
                bias=attention_bias,
                upcast_attention=upcast_attention,
            )
            nn.init.zeros_(self.attn_temporal.to_out[0].weight.data)  # initialize as an identity function
            self.norm_temporal = (
                AdaLayerNorm(dim, num_embeds_ada_norm) if self.use_ada_layer_norm else nn.LayerNorm(dim)
            )
        else:
            self.attn_temporal=None

        # efficient_attention_backward_cutlass is not implemented for large channels
        self.use_xformers = (dim <= 320) or "3090" not in torch.cuda.get_device_name(0)

        # 4. Feed-forward
        self.ff = FeedForward(dim, dropout=dropout, activation_fn=activation_fn)
        self.norm3 = nn.LayerNorm(dim)

    def set_use_memory_efficient_attention_xformers(self, use_memory_efficient_attention_xformers: bool, attention_op: Optional[Callable] = None):
        if not is_xformers_available():
            print("Here is how to install it")
            raise ModuleNotFoundError(
                "Refer to https://github.com/facebookresearch/xformers for more information on how to install"
                " xformers",
                name="xformers",
            )
        elif not torch.cuda.is_available():
            raise ValueError(
                "torch.cuda.is_available() should be True but is False. xformers' memory efficient attention is only"
                " available for GPU "
            )
        else:
            try:
                # Make sure we can run the memory efficient attention
                if use_memory_efficient_attention_xformers is True:
                    
                    _ = xformers.ops.memory_efficient_attention(
                        torch.randn((1, 2, 40), device="cuda"),
                        torch.randn((1, 2, 40), device="cuda"),
                        torch.randn((1, 2, 40), device="cuda"),
                    )
                else:
                    
                    pass
            except Exception as e:
                raise e
            # self.attn1._use_memory_efficient_attention_xformers = use_memory_efficient_attention_xformers
            # self.attn2._use_memory_efficient_attention_xformers = use_memory_efficient_attention_xformers
            self.attn1._use_memory_efficient_attention_xformers = use_memory_efficient_attention_xformers 
            self.attn2._use_memory_efficient_attention_xformers = use_memory_efficient_attention_xformers 
            # self.attn_temporal._use_memory_efficient_attention_xformers = (
            #     use_memory_efficient_attention_xformers
            # ),  # FIXME: enabling this raises CUDA ERROR. Gotta dig in.

    def forward(
        self,
        hidden_states,
        encoder_hidden_states=None,
        timestep=None,
        attention_mask=None,
        clip_length=None,
        **kwargs,
    ):
        # 1. Self-Attention
        norm_hidden_states = (
            self.norm1(hidden_states, timestep) if self.use_ada_layer_norm else self.norm1(hidden_states)
        )

        # Update kwargs instead of creating a new dictionary
        kwargs.update(
            hidden_states=norm_hidden_states,
            attention_mask=attention_mask,
        )
        # kwargs = dict(
        #     hidden_states=norm_hidden_states,
        #     attention_mask=attention_mask,
        # )
        if self.only_cross_attention:
            kwargs.update(encoder_hidden_states=encoder_hidden_states)
        if self.use_sparse_causal_attention:
            kwargs.update(clip_length=clip_length)
        if 'SparseCausalAttention_index' in self.model_config.keys():
            kwargs.update(SparseCausalAttention_index = self.model_config['SparseCausalAttention_index'])
        
        hidden_states = hidden_states + self.attn1(**kwargs)

        if clip_length is not None and self.temporal_attention_position == "after_spatial":
            hidden_states = self.apply_temporal_attention(hidden_states, timestep, clip_length)
        # print('hidden_states after 1 self attention',hidden_states.shape)
        if self.attn2 is not None:
            # 2. Cross-Attention
            norm_hidden_states = (
                self.norm2(hidden_states, timestep)
                if self.use_ada_layer_norm
                else self.norm2(hidden_states)
            )
            # print('norm_hidden_states',norm_hidden_states.shape)
            hidden_states = (
                self.attn2(
                    norm_hidden_states, # [16, 4096, 320]
                    encoder_hidden_states=encoder_hidden_states, # [1, 77, 768]
                    attention_mask=attention_mask,
                )
                + hidden_states
            )

        if clip_length is not None and self.temporal_attention_position == "after_cross" and self.attn_temporal is not None:
            hidden_states = self.apply_temporal_attention(hidden_states, timestep, clip_length)

        # 3. Feed-forward
        hidden_states = self.ff(self.norm3(hidden_states)) + hidden_states

        # if clip_length is not None and self.temporal_attention_position == "after_feedforward" and self.attn_temporal is not None:
        #     hidden_states = self.apply_temporal_attention(hidden_states, timestep, clip_length)

        return hidden_states

    def apply_temporal_attention(self, hidden_states, timestep, clip_length):
        d = hidden_states.shape[1]
        hidden_states = rearrange(hidden_states, "(b f) d c -> (b d) f c", f=clip_length)
        norm_hidden_states = (
            self.norm_temporal(hidden_states, timestep)
            if self.use_ada_layer_norm
            else self.norm_temporal(hidden_states)
        )
        hidden_states = self.attn_temporal(norm_hidden_states) + hidden_states
        hidden_states = rearrange(hidden_states, "(b d) f c -> (b f) d c", d=d)
        return hidden_states


def generate_dynamic_window_index():
    windows = [(0,5), (6,13), (14,21), (22,31), (32,39), (40,51), (52,58), (59,69)]
    video_index = []

    for start, end in windows:
        for i in range(start, end + 1):
            video_index.append(start)

    return video_index

def generate_fix_anchor_frames(clip_length, interval):
    # 创建锚点索引列表
    anchors = list(range(0, clip_length, interval))
    # 确保最后一个索引不超过clip_length
    if anchors[-1] >= clip_length:
        anchors[-1] = clip_length - 1
    # 生成每个锚点重复clip_length次的列表
    frame_index_list = [[anchor] * clip_length for anchor in anchors]
    return frame_index_list

def generate_random_anchor_frame_index(clip_length, interval):
    # 计算每个间隔的起始帧索引
    interval_starts = torch.arange(0, clip_length, interval)

    # 初始化锚点帧列表
    frame_index_list = []

    # 在每个间隔内随机选择一个帧索引
    for start in interval_starts:
        # 确保随机索引不超过clip_length
        end = min(start + interval, clip_length)
        anchor_frame = torch.randint(start, end, (1,)).item()

        # 生成重复clip_length次的锚点帧索引列表
        frame_index = [anchor_frame] * clip_length
        frame_index_list.append(frame_index)

    return frame_index_list

class SparseCausalAttention(CrossAttention):
    def forward(
        self,
        hidden_states,
        encoder_hidden_states=None,
        attention_mask=None,
        clip_length: int = None,
        SparseCausalAttention_index: list = [-1, 'first']   #list = ['anchor_interval8',0]     #list = [0]  #list = [-1, 'first','dynamic']
    ):
        if (
            self.added_kv_proj_dim is not None
            or encoder_hidden_states is not None
            or attention_mask is not None
        ):
            raise NotImplementedError

        if self.group_norm is not None:
            hidden_states = self.group_norm(hidden_states.transpose(1, 2)).transpose(1, 2)

        query = self.to_q(hidden_states)
        dim = query.shape[-1]
        query = self.reshape_heads_to_batch_dim(query)

        key = self.to_k(hidden_states)
        value = self.to_v(hidden_states)

        if clip_length is not None:
            key = rearrange(key, "(b f) d c -> b f d c", f=clip_length)
            value = rearrange(value, "(b f) d c -> b f d c", f=clip_length)


            #  *********************** Start of Spatial-temporal attention **********
            frame_index_list = []
            # print(f'SparseCausalAttention_index {str(SparseCausalAttention_index)}')
            if len(SparseCausalAttention_index) > 0:
                for index in SparseCausalAttention_index:
                    if isinstance(index, str):
                        if index == 'first':
                            frame_index = [0] * clip_length
                        if index == 'last':
                            frame_index = [clip_length-1] * clip_length
                        if (index == 'mid') or (index == 'middle'):
                            frame_index = [int(clip_length-1)//2] * clip_length
                        if index == "dynamic":
                            frame_index = generate_dynamic_window_index()
                        if index == "anchor_interval8":
                            frame_index = generate_fix_anchor_frames(clip_length,8)

                    else:
                        assert isinstance(index, int), 'relative index must be int'
                        frame_index = torch.arange(clip_length) + index
                        frame_index = frame_index.clip(0, clip_length-1)

                    if isinstance(frame_index[0], list):
                        frame_index_list = frame_index
                    else:
                        frame_index_list.append(frame_index)
                # print("frame_index_list",frame_index_list)
                # print("key before concat",key.shape)   #[1, frame, 4096, 320]
                key = torch.cat([   key[:, frame_index] for frame_index in frame_index_list
                                    ], dim=2)
                value = torch.cat([ value[:, frame_index] for frame_index in frame_index_list
                                    ], dim=2)

            #  *********************** End of Spatial-temporal attention **********
            key = rearrange(key, "b f d c -> (b f) d c", f=clip_length)
            value = rearrange(value, "b f d c -> (b f) d c", f=clip_length)
        
        
        key = self.reshape_heads_to_batch_dim(key)
        value = self.reshape_heads_to_batch_dim(value)
        
        
        if self._use_memory_efficient_attention_xformers:
            hidden_states = self._memory_efficient_attention_xformers(query, key, value, attention_mask)
            # Some versions of xformers return output in fp32, cast it back to the dtype of the input
            hidden_states = hidden_states.to(query.dtype)
        else:
            if self._slice_size is None or query.shape[0] // self._slice_size == 1:
                hidden_states = self._attention(query, key, value, attention_mask)
            else:
                hidden_states = self._sliced_attention(
                    query, key, value, hidden_states.shape[1], dim, attention_mask
                )

        # linear proj
        hidden_states = self.to_out[0](hidden_states)

        # dropout
        hidden_states = self.to_out[1](hidden_states)
        return hidden_states



class FullyFrameAttention(nn.Module):
    r"""
    A cross attention layer.

    Parameters:
        query_dim (`int`): The number of channels in the query.
        cross_attention_dim (`int`, *optional*):
            The number of channels in the encoder_hidden_states. If not given, defaults to `query_dim`.
        heads (`int`,  *optional*, defaults to 8): The number of heads to use for multi-head attention.
        dim_head (`int`,  *optional*, defaults to 64): The number of channels in each head.
        dropout (`float`, *optional*, defaults to 0.0): The dropout probability to use.
        bias (`bool`, *optional*, defaults to False):
            Set to `True` for the query, key, and value linear layers to contain a bias parameter.
    """

    def __init__(
        self,
        query_dim: int,
        cross_attention_dim: Optional[int] = None,
        heads: int = 8,
        dim_head: int = 64,
        dropout: float = 0.0,
        bias=False,
        upcast_attention: bool = False,
        upcast_softmax: bool = False,
        added_kv_proj_dim: Optional[int] = None,
        norm_num_groups: Optional[int] = None,
    ):
        super().__init__()
        inner_dim = dim_head * heads
        cross_attention_dim = cross_attention_dim if cross_attention_dim is not None else query_dim
        self.upcast_attention = upcast_attention
        self.upcast_softmax = upcast_softmax

        self.scale = dim_head**-0.5

        self.heads = heads
        # for slice_size > 0 the attention score computation
        # is split across the batch axis to save memory
        # You can set slice_size with `set_attention_slice`
        self.sliceable_head_dim = heads
        self._slice_size = heads
        # self._slice_size = None
        self._use_memory_efficient_attention_xformers = False
        self.added_kv_proj_dim = added_kv_proj_dim

        if norm_num_groups is not None:
            self.group_norm = nn.GroupNorm(num_channels=inner_dim, num_groups=norm_num_groups, eps=1e-5, affine=True)
        else:
            self.group_norm = None

        self.to_q = nn.Linear(query_dim, inner_dim, bias=bias)
        self.to_k = nn.Linear(cross_attention_dim, inner_dim, bias=bias)
        self.to_v = nn.Linear(cross_attention_dim, inner_dim, bias=bias)

        if self.added_kv_proj_dim is not None:
            self.add_k_proj = nn.Linear(added_kv_proj_dim, cross_attention_dim)
            self.add_v_proj = nn.Linear(added_kv_proj_dim, cross_attention_dim)

        self.to_out = nn.ModuleList([])
        self.to_out.append(nn.Linear(inner_dim, query_dim))
        self.to_out.append(nn.Dropout(dropout))

        self.q = None
        self.inject_q = None
        self.k = None
        self.inject_k = None

    def reshape_heads_to_batch_dim(self, tensor):
        batch_size, seq_len, dim = tensor.shape
        head_size = self.heads
        tensor = tensor.reshape(batch_size, seq_len, head_size, dim // head_size)
        tensor = tensor.permute(0, 2, 1, 3).reshape(batch_size * head_size, seq_len, dim // head_size)
        return tensor

    def reshape_batch_dim_to_heads(self, tensor):
        batch_size, seq_len, dim = tensor.shape
        head_size = self.heads
        tensor = tensor.reshape(batch_size // head_size, head_size, seq_len, dim)
        tensor = tensor.permute(0, 2, 1, 3).reshape(batch_size // head_size, seq_len, dim * head_size)
        return tensor

    def set_attention_slice(self, slice_size):
        if slice_size is not None and slice_size > self.sliceable_head_dim:
            raise ValueError(f"slice_size {slice_size} has to be smaller or equal to {self.sliceable_head_dim}.")

        self._slice_size = slice_size
    
    def _attention(self, query, key, value, attention_mask=None):
        if self.upcast_attention:
            query = query.float()
            key = key.float()

        attention_scores = torch.baddbmm(
            torch.empty(query.shape[0], query.shape[1], key.shape[1], dtype=query.dtype, device=query.device),
            query,
            key.transpose(-1, -2),
            beta=0,
            alpha=self.scale,
        )
        if attention_mask is not None:
            attention_scores = attention_scores + attention_mask

        if self.upcast_softmax:
            attention_scores = attention_scores.float()

        attention_probs = attention_scores.softmax(dim=-1)

        # cast back to the original dtype
        attention_probs = attention_probs.to(value.dtype)

        # compute attention output
        hidden_states = torch.bmm(attention_probs, value)

        # reshape hidden_states
        hidden_states = self.reshape_batch_dim_to_heads(hidden_states)
        return hidden_states

    def _sliced_attention(self, query, key, value, sequence_length, dim, attention_mask):

        batch_size_attention = query.shape[0]
        hidden_states = torch.zeros(
            (batch_size_attention, sequence_length, dim // self.heads), device=query.device, dtype=query.dtype
        )
        slice_size = self._slice_size if self._slice_size is not None else hidden_states.shape[0]
        for i in range(hidden_states.shape[0] // slice_size):
            start_idx = i * slice_size
            end_idx = (i + 1) * slice_size

            query_slice = query[start_idx:end_idx]
            key_slice = key[start_idx:end_idx]

            if self.upcast_attention:
                query_slice = query_slice.float()
                key_slice = key_slice.float()

            attn_slice = torch.baddbmm(
                torch.empty(slice_size, query.shape[1], key.shape[1], dtype=query_slice.dtype, device=query.device),
                query_slice,
                key_slice.transpose(-1, -2),
                beta=0,
                alpha=self.scale,
            )

            if attention_mask is not None:
                attn_slice = attn_slice + attention_mask[start_idx:end_idx]

            if self.upcast_softmax:
                attn_slice = attn_slice.float()

            attn_slice = attn_slice.softmax(dim=-1)

            # cast back to the original dtype
            attn_slice = attn_slice.to(value.dtype)
            attn_slice = torch.bmm(attn_slice, value[start_idx:end_idx])

            hidden_states[start_idx:end_idx] = attn_slice

        # reshape hidden_states
        hidden_states = self.reshape_batch_dim_to_heads(hidden_states)

        return hidden_states

    def reshape_heads_to_batch_dim3(self, tensor):
        batch_size1, batch_size2, seq_len, dim = tensor.shape
        head_size = self.heads
        tensor = tensor.reshape(batch_size1, batch_size2, seq_len, head_size, dim // head_size)
        tensor = tensor.permute(0, 3, 1, 2, 4)
        return tensor
    
    def _memory_efficient_attention_xformers(self, query, key, value, attention_mask):
        # TODO attention_mask
        query = query.contiguous()
        key = key.contiguous()
        value = value.contiguous()
        hidden_states = xformers.ops.memory_efficient_attention(query, key, value, attn_bias=attention_mask)
        hidden_states = self.reshape_batch_dim_to_heads(hidden_states)
        return hidden_states

    def forward(self, hidden_states, encoder_hidden_states=None, attention_mask=None, clip_length=None, inter_frame=False,**kwargs):
        batch_size, sequence_length, _ = hidden_states.shape

        encoder_hidden_states = encoder_hidden_states

        # h = w = int(math.sqrt(sequence_length))
        h = kwargs['height']
        w = kwargs['width']
        if self.group_norm is not None:
            hidden_states = self.group_norm(hidden_states.transpose(1, 2)).transpose(1, 2)


        query = self.to_q(hidden_states)  # (bf) x d(hw) x c

        self.q = query
        if self.inject_q is not None:
            query = self.inject_q
        dim = query.shape[-1]
        query_old = query.clone()
        dim = query.shape[-1]

        # All frames
        query = rearrange(query, "(b f) d c -> b (f d) c", f=clip_length)
        
        query = self.reshape_heads_to_batch_dim(query)

        if self.added_kv_proj_dim is not None:
            raise NotImplementedError

        encoder_hidden_states = encoder_hidden_states if encoder_hidden_states is not None else hidden_states
        key = self.to_k(encoder_hidden_states)
        self.k = key
        if self.inject_k is not None:
            key = self.inject_k
        key_old = key.clone()
        value = self.to_v(encoder_hidden_states)

        if inter_frame:
            key = rearrange(key, "(b f) d c -> b f d c", f=clip_length)[:, [0, -1]]
            value = rearrange(value, "(b f) d c -> b f d c", f=clip_length)[:, [0, -1]]
            key = rearrange(key, "b f d c -> b (f d) c",)
            value = rearrange(value, "b f d c -> b (f d) c")
        else:
            # All frames
            key = rearrange(key, "(b f) d c -> b (f d) c", f=clip_length)
            value = rearrange(value, "(b f) d c -> b (f d) c", f=clip_length)

        key = self.reshape_heads_to_batch_dim(key)
        value = self.reshape_heads_to_batch_dim(value)

        if attention_mask is not None:
            if attention_mask.shape[-1] != query.shape[1]:
                target_length = query.shape[1]
                attention_mask = F.pad(attention_mask, (0, target_length), value=0.0)
                attention_mask = attention_mask.repeat_interleave(self.heads, dim=0)

        # attention, what we cannot get enough of
        if self._use_memory_efficient_attention_xformers:
            hidden_states = self._memory_efficient_attention_xformers(query, key, value, attention_mask)
            # Some versions of xformers return output in fp32, cast it back to the dtype of the input
            hidden_states = hidden_states.to(query.dtype)
        else:
            if self._slice_size is None or query.shape[0] // self._slice_size == 1:
                hidden_states = self._attention(query, key, value, attention_mask)
            else:
                hidden_states = self._sliced_attention(query, key, value, sequence_length, dim, attention_mask)
        
        
        if [h,w] in kwargs['flatten_res']:

            hidden_states = rearrange(hidden_states, "b (f d) c -> (b f) d c", f=clip_length)
            if self.group_norm is not None:
                hidden_states = self.group_norm(hidden_states.transpose(1, 2)).transpose(1, 2)

            if kwargs["old_qk"] == 1:
                query = query_old
                key = key_old
            else:
                query = hidden_states
                key = hidden_states
            value = hidden_states

            traj = kwargs["traj"]
            traj = rearrange(traj, '(f n) l d -> f n l d', f=clip_length, n=sequence_length)
            mask = rearrange(kwargs["mask"], '(f n) l -> f n l', f=clip_length, n=sequence_length)
            mask = torch.cat([mask[:, :, 0].unsqueeze(-1), mask[:, :, -clip_length+1:]], dim=-1)

            # print('traj',traj.shape)
            # print('mask',mask.shape)

            traj_key_sequence_inds = torch.cat([traj[:, :, 0, :].unsqueeze(-2), traj[:, :, -clip_length+1:, :]], dim=-2)
            t_inds = traj_key_sequence_inds[:, :, :, 0]
            x_inds = traj_key_sequence_inds[:, :, :, 1]
            y_inds = traj_key_sequence_inds[:, :, :, 2]

            query_tempo = query.unsqueeze(-2)
            _key = rearrange(key, '(b f) (h w) d -> b f h w d', b=int(batch_size/clip_length), f=clip_length, h=h, w=w)
            _value = rearrange(value, '(b f) (h w) d -> b f h w d', b=int(batch_size/clip_length), f=clip_length, h=h, w=w)
            key_tempo = _key[:, t_inds, x_inds, y_inds]
            value_tempo = _value[:, t_inds, x_inds, y_inds]
            key_tempo = rearrange(key_tempo, 'b f n l d -> (b f) n l d')
            value_tempo = rearrange(value_tempo, 'b f n l d -> (b f) n l d')

            mask = rearrange(torch.stack([mask, mask]),  'b f n l -> (b f) n l')
            mask = mask[:,None].repeat(1, self.heads, 1, 1).unsqueeze(-2)
            attn_bias = torch.zeros_like(mask, dtype=key_tempo.dtype) # regular zeros_like
            attn_bias[~mask] = -torch.inf

            # print('attn_bias',attn_bias.shape)
            # print('query_tempo',query_tempo.shape)
            # print('key_tempo',key_tempo.shape)

            # flow attention
            query_tempo = self.reshape_heads_to_batch_dim3(query_tempo)
            key_tempo = self.reshape_heads_to_batch_dim3(key_tempo)
            value_tempo = self.reshape_heads_to_batch_dim3(value_tempo)

            attn_matrix2 = query_tempo @ key_tempo.transpose(-2, -1) / math.sqrt(query_tempo.size(-1)) + attn_bias
            attn_matrix2 = F.softmax(attn_matrix2, dim=-1)
            out = (attn_matrix2@value_tempo).squeeze(-2)

            hidden_states = rearrange(out,'(b f) k (h w) d -> b (f h w) (k d)', b=int(batch_size/clip_length), f=clip_length, h=h, w=w)
        
        
        # linear proj
        hidden_states = self.to_out[0](hidden_states)

        # dropout
        hidden_states = self.to_out[1](hidden_states)

        # All frames
        hidden_states = rearrange(hidden_states, "b (f d) c -> (b f) d c", f=clip_length)
        return hidden_states