# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0
#
# 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.

# Copyright 2024-2025 The Alibaba Wan Team Authors. All rights reserved.

import time
from contextlib import nullcontext

import torch
import torch.nn as nn
import torch.nn.functional as F
from einops import rearrange

from cosmos_predict2._src.imaginaire.utils import log
from cosmos_predict2._src.imaginaire.utils.distributed import get_rank, sync_model_states
from cosmos_predict2._src.imaginaire.utils.easy_io import easy_io
from cosmos_predict2._src.predict2.tokenizers.interface import VideoTokenizerInterface
from cosmos_predict2._src.predict2.utils.tokenizer_benchmarking import BenchmarkTimes

__all__ = [
    "WanVAE",
]

CACHE_T = 2


class CausalConv3d(nn.Conv3d):
    """
    Causal 3d convolusion.
    """

    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
        self._padding = (
            self.padding[2],
            self.padding[2],
            self.padding[1],
            self.padding[1],
            2 * self.padding[0],
            0,
        )
        self.padding = (0, 0, 0)

    def forward(self, x, cache_x=None):
        padding = list(self._padding)
        if cache_x is not None and self._padding[4] > 0:
            cache_x = cache_x.to(x.device)
            x = torch.cat([cache_x, x], dim=2)
            padding[4] -= cache_x.shape[2]
        x = F.pad(x, padding)

        return super().forward(x)


class RMS_norm(nn.Module):
    def __init__(self, dim, channel_first=True, images=True, bias=False):
        super().__init__()
        broadcastable_dims = (1, 1, 1) if not images else (1, 1)
        shape = (dim, *broadcastable_dims) if channel_first else (dim,)

        self.channel_first = channel_first
        self.scale = dim**0.5
        self.gamma = nn.Parameter(torch.ones(shape))
        self.bias = nn.Parameter(torch.zeros(shape)) if bias else 0.0

    def forward(self, x):
        return F.normalize(x, dim=(1 if self.channel_first else -1)) * self.scale * self.gamma + self.bias


class Upsample(nn.Upsample):
    def forward(self, x):
        """
        Fix bfloat16 support for nearest neighbor interpolation.
        """
        return super().forward(x.float()).type_as(x)


class Resample(nn.Module):
    def __init__(self, dim, mode):
        assert mode in (
            "none",
            "upsample2d",
            "upsample3d",
            "downsample2d",
            "downsample3d",
        )
        super().__init__()
        self.dim = dim
        self.mode = mode

        # layers
        if mode == "upsample2d":
            self.resample = nn.Sequential(
                Upsample(scale_factor=(2.0, 2.0), mode="nearest-exact"),
                nn.Conv2d(dim, dim, 3, padding=1),
            )
        elif mode == "upsample3d":
            self.resample = nn.Sequential(
                Upsample(scale_factor=(2.0, 2.0), mode="nearest-exact"),
                nn.Conv2d(dim, dim, 3, padding=1),
            )
            self.time_conv = CausalConv3d(dim, dim * 2, (3, 1, 1), padding=(1, 0, 0))
        elif mode == "downsample2d":
            self.resample = nn.Sequential(nn.ZeroPad2d((0, 1, 0, 1)), nn.Conv2d(dim, dim, 3, stride=(2, 2)))
        elif mode == "downsample3d":
            self.resample = nn.Sequential(nn.ZeroPad2d((0, 1, 0, 1)), nn.Conv2d(dim, dim, 3, stride=(2, 2)))
            self.time_conv = CausalConv3d(dim, dim, (3, 1, 1), stride=(2, 1, 1), padding=(0, 0, 0))
        else:
            self.resample = nn.Identity()

    def forward(self, x, feat_cache=None, feat_idx=[0]):
        b, c, t, h, w = x.size()
        if self.mode == "upsample3d":
            if feat_cache is not None:
                idx = feat_idx[0]
                if feat_cache[idx] is None:
                    feat_cache[idx] = "Rep"
                    feat_idx[0] += 1
                else:
                    cache_x = x[:, :, -CACHE_T:, :, :].clone()
                    if cache_x.shape[2] < 2 and feat_cache[idx] is not None and feat_cache[idx] != "Rep":
                        # cache last frame of last two chunk
                        cache_x = torch.cat(
                            [
                                feat_cache[idx][:, :, -1, :, :].unsqueeze(2).to(cache_x.device),
                                cache_x,
                            ],
                            dim=2,
                        )
                    if cache_x.shape[2] < 2 and feat_cache[idx] is not None and feat_cache[idx] == "Rep":
                        cache_x = torch.cat(
                            [torch.zeros_like(cache_x).to(cache_x.device), cache_x],
                            dim=2,
                        )
                    if feat_cache[idx] == "Rep":
                        x = self.time_conv(x)
                    else:
                        x = self.time_conv(x, feat_cache[idx])
                    feat_cache[idx] = cache_x
                    feat_idx[0] += 1
                    x = x.reshape(b, 2, c, t, h, w)
                    x = torch.stack((x[:, 0, :, :, :, :], x[:, 1, :, :, :, :]), 3)
                    x = x.reshape(b, c, t * 2, h, w)
        t = x.shape[2]
        x = rearrange(x, "b c t h w -> (b t) c h w")
        x = self.resample(x)
        x = rearrange(x, "(b t) c h w -> b c t h w", t=t)

        if self.mode == "downsample3d":
            if feat_cache is not None:
                idx = feat_idx[0]
                if feat_cache[idx] is None:
                    feat_cache[idx] = x.clone()
                    feat_idx[0] += 1
                else:
                    cache_x = x[:, :, -1:, :, :].clone()
                    x = self.time_conv(torch.cat([feat_cache[idx][:, :, -1:, :, :], x], 2))
                    feat_cache[idx] = cache_x
                    feat_idx[0] += 1
        return x


class ResidualBlock(nn.Module):
    def __init__(self, in_dim, out_dim, dropout=0.0):
        super().__init__()
        self.in_dim = in_dim
        self.out_dim = out_dim

        # layers
        self.residual = nn.Sequential(
            RMS_norm(in_dim, images=False),
            nn.SiLU(),
            CausalConv3d(in_dim, out_dim, 3, padding=1),
            RMS_norm(out_dim, images=False),
            nn.SiLU(),
            nn.Dropout(dropout),
            CausalConv3d(out_dim, out_dim, 3, padding=1),
        )
        self.shortcut = CausalConv3d(in_dim, out_dim, 1) if in_dim != out_dim else nn.Identity()

    def forward(self, x, feat_cache=None, feat_idx=[0]):
        h = self.shortcut(x)
        for layer in self.residual:
            if isinstance(layer, CausalConv3d) and feat_cache is not None:
                idx = feat_idx[0]
                cache_x = x[:, :, -CACHE_T:, :, :].clone()
                if cache_x.shape[2] < 2 and feat_cache[idx] is not None:
                    # cache last frame of last two chunk
                    cache_x = torch.cat(
                        [
                            feat_cache[idx][:, :, -1, :, :].unsqueeze(2).to(cache_x.device),
                            cache_x,
                        ],
                        dim=2,
                    )
                x = layer(x, feat_cache[idx])
                feat_cache[idx] = cache_x
                feat_idx[0] += 1
            else:
                x = layer(x)
        return x + h


class AttentionBlock(nn.Module):
    """
    Causal self-attention with a single head.
    """

    def __init__(self, dim):
        super().__init__()
        self.dim = dim

        # layers
        self.norm = RMS_norm(dim)
        self.to_qkv = nn.Conv2d(dim, dim * 3, 1)
        self.proj = nn.Conv2d(dim, dim, 1)

        # zero out the last layer params
        nn.init.zeros_(self.proj.weight)

    def forward(self, x):
        identity = x
        b, c, t, h, w = x.size()
        x = rearrange(x, "b c t h w -> (b t) c h w")
        x = self.norm(x)
        # compute query, key, value
        q, k, v = self.to_qkv(x).reshape(b * t, 1, c * 3, -1).permute(0, 1, 3, 2).contiguous().chunk(3, dim=-1)

        # apply attention
        x = F.scaled_dot_product_attention(
            q,
            k,
            v,
        )
        x = x.squeeze(1).permute(0, 2, 1).reshape(b * t, c, h, w)

        # output
        x = self.proj(x)
        x = rearrange(x, "(b t) c h w-> b c t h w", t=t)
        return x + identity


def patchify(x, patch_size):
    if patch_size == 1:
        return x
    if x.dim() == 4:
        x = rearrange(x, "b c (h q) (w r) -> b (c r q) h w", q=patch_size, r=patch_size)
    elif x.dim() == 5:
        x = rearrange(
            x,
            "b c f (h q) (w r) -> b (c r q) f h w",
            q=patch_size,
            r=patch_size,
        )
    else:
        raise ValueError(f"Invalid input shape: {x.shape}")

    return x


def unpatchify(x, patch_size):
    if patch_size == 1:
        return x

    if x.dim() == 4:
        x = rearrange(x, "b (c r q) h w -> b c (h q) (w r)", q=patch_size, r=patch_size)
    elif x.dim() == 5:
        x = rearrange(
            x,
            "b (c r q) f h w -> b c f (h q) (w r)",
            q=patch_size,
            r=patch_size,
        )
    return x


class AvgDown3D(nn.Module):
    def __init__(
        self,
        in_channels,
        out_channels,
        factor_t,
        factor_s=1,
    ):
        super().__init__()
        self.in_channels = in_channels
        self.out_channels = out_channels
        self.factor_t = factor_t
        self.factor_s = factor_s
        self.factor = self.factor_t * self.factor_s * self.factor_s

        assert in_channels * self.factor % out_channels == 0
        self.group_size = in_channels * self.factor // out_channels

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        pad_t = (self.factor_t - x.shape[2] % self.factor_t) % self.factor_t
        pad = (0, 0, 0, 0, pad_t, 0)
        x = F.pad(x, pad)
        B, C, T, H, W = x.shape
        x = x.view(
            B,
            C,
            T // self.factor_t,
            self.factor_t,
            H // self.factor_s,
            self.factor_s,
            W // self.factor_s,
            self.factor_s,
        )
        x = x.permute(0, 1, 3, 5, 7, 2, 4, 6).contiguous()
        x = x.view(
            B,
            C * self.factor,
            T // self.factor_t,
            H // self.factor_s,
            W // self.factor_s,
        )
        x = x.view(
            B,
            self.out_channels,
            self.group_size,
            T // self.factor_t,
            H // self.factor_s,
            W // self.factor_s,
        )
        x = x.mean(dim=2)
        return x


class DupUp3D(nn.Module):
    def __init__(
        self,
        in_channels: int,
        out_channels: int,
        factor_t,
        factor_s=1,
    ):
        super().__init__()
        self.in_channels = in_channels
        self.out_channels = out_channels

        self.factor_t = factor_t
        self.factor_s = factor_s
        self.factor = self.factor_t * self.factor_s * self.factor_s

        assert out_channels * self.factor % in_channels == 0
        self.repeats = out_channels * self.factor // in_channels

    def forward(self, x: torch.Tensor, first_chunk=False) -> torch.Tensor:
        x = x.repeat_interleave(self.repeats, dim=1)
        x = x.view(
            x.size(0),
            self.out_channels,
            self.factor_t,
            self.factor_s,
            self.factor_s,
            x.size(2),
            x.size(3),
            x.size(4),
        )
        x = x.permute(0, 1, 5, 2, 6, 3, 7, 4).contiguous()
        x = x.view(
            x.size(0),
            self.out_channels,
            x.size(2) * self.factor_t,
            x.size(4) * self.factor_s,
            x.size(6) * self.factor_s,
        )
        if first_chunk:
            x = x[:, :, self.factor_t - 1 :, :, :]
        return x


class Down_ResidualBlock(nn.Module):
    def __init__(self, in_dim, out_dim, dropout, mult, temperal_downsample=False, down_flag=False):
        super().__init__()

        # Shortcut path with downsample
        self.avg_shortcut = AvgDown3D(
            in_dim,
            out_dim,
            factor_t=2 if temperal_downsample else 1,
            factor_s=2 if down_flag else 1,
        )

        # Main path with residual blocks and downsample
        downsamples = []
        for _ in range(mult):
            downsamples.append(ResidualBlock(in_dim, out_dim, dropout))
            in_dim = out_dim

        # Add the final downsample block
        if down_flag:
            mode = "downsample3d" if temperal_downsample else "downsample2d"
            downsamples.append(Resample(out_dim, mode=mode))

        self.downsamples = nn.Sequential(*downsamples)

    def forward(self, x, feat_cache=None, feat_idx=[0]):
        x_copy = x.clone()
        for module in self.downsamples:
            x = module(x, feat_cache, feat_idx)

        return x + self.avg_shortcut(x_copy)


class Up_ResidualBlock(nn.Module):
    def __init__(self, in_dim, out_dim, dropout, mult, temperal_upsample=False, up_flag=False):
        super().__init__()
        # Shortcut path with upsample
        if up_flag:
            self.avg_shortcut = DupUp3D(
                in_dim,
                out_dim,
                factor_t=2 if temperal_upsample else 1,
                factor_s=2 if up_flag else 1,
            )
        else:
            self.avg_shortcut = None

        # Main path with residual blocks and upsample
        upsamples = []
        for _ in range(mult):
            upsamples.append(ResidualBlock(in_dim, out_dim, dropout))
            in_dim = out_dim

        # Add the final upsample block
        if up_flag:
            mode = "upsample3d" if temperal_upsample else "upsample2d"
            upsamples.append(Resample(out_dim, mode=mode))

        self.upsamples = nn.Sequential(*upsamples)

    def forward(self, x, feat_cache=None, feat_idx=[0], first_chunk=False):
        x_main = x.clone()
        for module in self.upsamples:
            x_main = module(x_main, feat_cache, feat_idx)
        if self.avg_shortcut is not None:
            x_shortcut = self.avg_shortcut(x, first_chunk)
            return x_main + x_shortcut
        else:
            return x_main


class Encoder3d(nn.Module):
    def __init__(
        self,
        dim=128,
        z_dim=4,
        dim_mult=[1, 2, 4, 4],
        num_res_blocks=2,
        attn_scales=[],
        temperal_downsample=[False, True, True],
        dropout=0.0,
    ):
        super().__init__()
        self.dim = dim
        self.z_dim = z_dim
        self.dim_mult = dim_mult
        self.num_res_blocks = num_res_blocks
        self.attn_scales = attn_scales
        self.temperal_downsample = temperal_downsample

        # dimensions
        dims = [dim * u for u in [1] + dim_mult]
        scale = 1.0

        # init block
        self.conv1 = CausalConv3d(12, dims[0], 3, padding=1)

        # downsample blocks
        downsamples = []
        for i, (in_dim, out_dim) in enumerate(zip(dims[:-1], dims[1:])):
            t_down_flag = temperal_downsample[i] if i < len(temperal_downsample) else False
            downsamples.append(
                Down_ResidualBlock(
                    in_dim=in_dim,
                    out_dim=out_dim,
                    dropout=dropout,
                    mult=num_res_blocks,
                    temperal_downsample=t_down_flag,
                    down_flag=i != len(dim_mult) - 1,
                )
            )
            scale /= 2.0
        self.downsamples = nn.Sequential(*downsamples)

        # middle blocks
        self.middle = nn.Sequential(
            ResidualBlock(out_dim, out_dim, dropout),
            AttentionBlock(out_dim),
            ResidualBlock(out_dim, out_dim, dropout),
        )

        # output blocks
        self.head = nn.Sequential(
            RMS_norm(out_dim, images=False),
            nn.SiLU(),
            CausalConv3d(out_dim, z_dim, 3, padding=1),
        )

    def forward(self, x, feat_cache=None, feat_idx=[0]):
        if feat_cache is not None:
            idx = feat_idx[0]
            cache_x = x[:, :, -CACHE_T:, :, :].clone()
            if cache_x.shape[2] < 2 and feat_cache[idx] is not None:
                cache_x = torch.cat(
                    [
                        feat_cache[idx][:, :, -1, :, :].unsqueeze(2).to(cache_x.device),
                        cache_x,
                    ],
                    dim=2,
                )
            x = self.conv1(x, feat_cache[idx])
            feat_cache[idx] = cache_x
            feat_idx[0] += 1
        else:
            x = self.conv1(x)

        # downsamples
        for layer in self.downsamples:
            if feat_cache is not None:
                x = layer(x, feat_cache, feat_idx)
            else:
                x = layer(x)

        # middle
        for layer in self.middle:
            if isinstance(layer, ResidualBlock) and feat_cache is not None:
                x = layer(x, feat_cache, feat_idx)
            else:
                x = layer(x)

        # head
        for layer in self.head:
            if isinstance(layer, CausalConv3d) and feat_cache is not None:
                idx = feat_idx[0]
                cache_x = x[:, :, -CACHE_T:, :, :].clone()
                if cache_x.shape[2] < 2 and feat_cache[idx] is not None:
                    cache_x = torch.cat(
                        [
                            feat_cache[idx][:, :, -1, :, :].unsqueeze(2).to(cache_x.device),
                            cache_x,
                        ],
                        dim=2,
                    )
                x = layer(x, feat_cache[idx])
                feat_cache[idx] = cache_x
                feat_idx[0] += 1
            else:
                x = layer(x)

        return x


class Decoder3d(nn.Module):
    def __init__(
        self,
        dim=128,
        z_dim=4,
        dim_mult=[1, 2, 4, 4],
        num_res_blocks=2,
        attn_scales=[],
        temperal_upsample=[True, True, False],
        dropout=0.0,
    ):
        super().__init__()
        self.dim = dim
        self.z_dim = z_dim
        self.dim_mult = dim_mult
        self.num_res_blocks = num_res_blocks
        self.attn_scales = attn_scales
        self.temperal_upsample = temperal_upsample

        # dimensions
        dims = [dim * u for u in [dim_mult[-1]] + dim_mult[::-1]]
        scale = 1.0 / 2 ** (len(dim_mult) - 2)
        # init block
        self.conv1 = CausalConv3d(z_dim, dims[0], 3, padding=1)

        # middle blocks
        self.middle = nn.Sequential(
            ResidualBlock(dims[0], dims[0], dropout),
            AttentionBlock(dims[0]),
            ResidualBlock(dims[0], dims[0], dropout),
        )

        # upsample blocks
        upsamples = []
        for i, (in_dim, out_dim) in enumerate(zip(dims[:-1], dims[1:])):
            t_up_flag = temperal_upsample[i] if i < len(temperal_upsample) else False
            upsamples.append(
                Up_ResidualBlock(
                    in_dim=in_dim,
                    out_dim=out_dim,
                    dropout=dropout,
                    mult=num_res_blocks + 1,
                    temperal_upsample=t_up_flag,
                    up_flag=i != len(dim_mult) - 1,
                )
            )
        self.upsamples = nn.Sequential(*upsamples)

        # output blocks
        self.head = nn.Sequential(
            RMS_norm(out_dim, images=False),
            nn.SiLU(),
            CausalConv3d(out_dim, 12, 3, padding=1),
        )

    def forward(self, x, feat_cache=None, feat_idx=[0], first_chunk=False):
        if feat_cache is not None:
            idx = feat_idx[0]
            cache_x = x[:, :, -CACHE_T:, :, :].clone()
            if cache_x.shape[2] < 2 and feat_cache[idx] is not None:
                cache_x = torch.cat(
                    [
                        feat_cache[idx][:, :, -1, :, :].unsqueeze(2).to(cache_x.device),
                        cache_x,
                    ],
                    dim=2,
                )
            x = self.conv1(x, feat_cache[idx])
            feat_cache[idx] = cache_x
            feat_idx[0] += 1
        else:
            x = self.conv1(x)

        for layer in self.middle:
            if isinstance(layer, ResidualBlock) and feat_cache is not None:
                x = layer(x, feat_cache, feat_idx)
            else:
                x = layer(x)

        # upsamples
        for layer in self.upsamples:
            if feat_cache is not None:
                x = layer(x, feat_cache, feat_idx, first_chunk)
            else:
                x = layer(x)

        # head
        for layer in self.head:
            if isinstance(layer, CausalConv3d) and feat_cache is not None:
                idx = feat_idx[0]
                cache_x = x[:, :, -CACHE_T:, :, :].clone()
                if cache_x.shape[2] < 2 and feat_cache[idx] is not None:
                    cache_x = torch.cat(
                        [
                            feat_cache[idx][:, :, -1, :, :].unsqueeze(2).to(cache_x.device),
                            cache_x,
                        ],
                        dim=2,
                    )
                x = layer(x, feat_cache[idx])
                feat_cache[idx] = cache_x
                feat_idx[0] += 1
            else:
                x = layer(x)
        return x


def count_conv3d(model):
    count = 0
    for m in model.modules():
        if isinstance(m, CausalConv3d):
            count += 1
    return count


class WanVAE_(nn.Module):
    def __init__(
        self,
        dim=160,
        dec_dim=256,
        z_dim=48,
        dim_mult=[1, 2, 4, 4],
        num_res_blocks=2,
        attn_scales=[],
        temperal_downsample=[False, True, True],
        dropout=0.0,
        temporal_window=4,
    ):
        super().__init__()
        self.dim = dim
        self.z_dim = z_dim
        self.dim_mult = dim_mult
        self.num_res_blocks = num_res_blocks
        self.attn_scales = attn_scales
        self.temperal_downsample = temperal_downsample
        self.temperal_upsample = temperal_downsample[::-1]
        self.temporal_window = temporal_window

        # modules
        self.encoder = Encoder3d(
            dim,
            z_dim * 2,
            dim_mult,
            num_res_blocks,
            attn_scales,
            self.temperal_downsample,
            dropout,
        )
        self.conv1 = CausalConv3d(z_dim * 2, z_dim * 2, 1)
        self.conv2 = CausalConv3d(z_dim, z_dim, 1)
        self.decoder = Decoder3d(
            dec_dim,
            z_dim,
            dim_mult,
            num_res_blocks,
            attn_scales,
            self.temperal_upsample,
            dropout,
        )

    def forward(self, x, scale=[0, 1]):
        mu = self.encode(x, scale)
        x_recon = self.decode(mu, scale)
        return x_recon, mu

    def encode(self, x, scale):
        self.clear_cache()
        x = patchify(x, patch_size=2)
        t = x.shape[2]
        iter_ = 1 + (t - 1) // self.temporal_window
        for i in range(iter_):
            self._enc_conv_idx = [0]
            if i == 0:
                out = self.encoder(
                    x[:, :, :1, :, :],
                    feat_cache=self._enc_feat_map,
                    feat_idx=self._enc_conv_idx,
                )
            else:
                out_ = self.encoder(
                    x[:, :, 1 + self.temporal_window * (i - 1) : 1 + self.temporal_window * i, :, :],
                    feat_cache=self._enc_feat_map,
                    feat_idx=self._enc_conv_idx,
                )
                out = torch.cat([out, out_], 2)
        if (t - 1) % self.temporal_window:
            self._enc_conv_idx = [0]
            out_ = self.encoder(
                x[:, :, 1 + self.temporal_window * (iter_ - 1) :, :, :],
                feat_cache=self._enc_feat_map,
                feat_idx=self._enc_conv_idx,
            )
            out = torch.cat([out, out_], 2)
        mu, log_var = self.conv1(out).chunk(2, dim=1)
        if isinstance(scale[0], torch.Tensor):
            mu = (mu - scale[0].view(1, self.z_dim, 1, 1, 1)) * scale[1].view(1, self.z_dim, 1, 1, 1)
        else:
            mu = (mu - scale[0]) * scale[1]
        self.clear_cache()
        return mu

    @torch.compiler.disable
    def _i0_encode(self, x):
        """
        If enabled torch.compile uses significantly more memory for this step, so we disable it
        """
        out = self.encoder(x[:, :, :1, :, :], feat_cache=self._enc_feat_map, feat_idx=self._enc_conv_idx)
        return out

    def decode(self, z, scale):
        self.clear_cache()
        if isinstance(scale[0], torch.Tensor):
            z = z / scale[1].view(1, self.z_dim, 1, 1, 1) + scale[0].view(1, self.z_dim, 1, 1, 1)
        else:
            z = z / scale[1] + scale[0]
        iter_ = z.shape[2]
        x = self.conv2(z)
        for i in range(iter_):
            self._conv_idx = [0]
            if i == 0:
                out = self.decoder(
                    x[:, :, i : i + 1, :, :],
                    feat_cache=self._feat_map,
                    feat_idx=self._conv_idx,
                    first_chunk=True,
                )
            else:
                out_ = self.decoder(
                    x[:, :, i : i + 1, :, :],
                    feat_cache=self._feat_map,
                    feat_idx=self._conv_idx,
                )
                out = torch.cat([out, out_], 2)
        out = unpatchify(out, patch_size=2)
        self.clear_cache()
        return out

    def reparameterize(self, mu, log_var):
        std = torch.exp(0.5 * log_var)
        eps = torch.randn_like(std)
        return eps * std + mu

    def sample(self, imgs, deterministic=False):
        mu, log_var = self.encode(imgs)
        if deterministic:
            return mu
        std = torch.exp(0.5 * log_var.clamp(-30.0, 20.0))
        return mu + std * torch.randn_like(std)

    def clear_cache(self):
        self._conv_num = count_conv3d(self.decoder)
        self._conv_idx = [0]
        self._feat_map = [None] * self._conv_num
        # cache encode
        self._enc_conv_num = count_conv3d(self.encoder)
        self._enc_conv_idx = [0]
        self._enc_feat_map = [None] * self._enc_conv_num


def _video_vae(
    pretrained_path=None,
    device="cpu",
    s3_credential_path: str = "credentials/s3_training.secret",
    **kwargs,
):
    """
    Autoencoder3d adapted from Wan 2.2.
    """
    # params
    cfg = dict(
        temperal_downsample=[False, True, True],
    )
    cfg.update(**kwargs)

    # init model
    with torch.device("meta"):
        model = WanVAE_(**cfg)

    if pretrained_path is None:
        model.to_empty(device=device)
    else:
        if get_rank() == 0:
            if pretrained_path.startswith("s3://"):
                backend_key = "wan2pt2_vae"
                easy_io.set_s3_backend(
                    key=backend_key,
                    backend_args={
                        "backend": "s3",
                        "s3_credential_path": s3_credential_path,
                    },
                )
            else:
                backend_key = None

            ckpt = easy_io.load(
                pretrained_path,
                backend_key=backend_key,
                map_location=device,
            )

            # load checkpoint
            log.info(f"loading {pretrained_path}")
            model.load_state_dict(ckpt, assign=True)
        else:
            model.to_empty(device=device)
    sync_model_states(model)

    return model


class WanVAE:
    def __init__(
        self,
        z_dim=48,
        vae_pth="s3://bucket/cosmos_diffusion_v2/pretrain_weights/tokenizer/wan2pt2/Wan2.2_VAE.pth",
        s3_credential_path: str = "credentials/s3_training.secret",
        dtype=torch.bfloat16,
        device="cuda",
        is_amp=True,
        benchmark: bool = False,
        temporal_window: int = 4,
    ):
        self.dtype = dtype
        self.device = device
        self.benchmark = benchmark
        self.temporal_window = temporal_window

        # Wan 2.2 mean and std values (48 dimensions)
        mean = [
            -0.2289,
            -0.0052,
            -0.1323,
            -0.2339,
            -0.2799,
            0.0174,
            0.1838,
            0.1557,
            -0.1382,
            0.0542,
            0.2813,
            0.0891,
            0.1570,
            -0.0098,
            0.0375,
            -0.1825,
            -0.2246,
            -0.1207,
            -0.0698,
            0.5109,
            0.2665,
            -0.2108,
            -0.2158,
            0.2502,
            -0.2055,
            -0.0322,
            0.1109,
            0.1567,
            -0.0729,
            0.0899,
            -0.2799,
            -0.1230,
            -0.0313,
            -0.1649,
            0.0117,
            0.0723,
            -0.2839,
            -0.2083,
            -0.0520,
            0.3748,
            0.0152,
            0.1957,
            0.1433,
            -0.2944,
            0.3573,
            -0.0548,
            -0.1681,
            -0.0667,
        ]
        std = [
            0.4765,
            1.0364,
            0.4514,
            1.1677,
            0.5313,
            0.4990,
            0.4818,
            0.5013,
            0.8158,
            1.0344,
            0.5894,
            1.0901,
            0.6885,
            0.6165,
            0.8454,
            0.4978,
            0.5759,
            0.3523,
            0.7135,
            0.6804,
            0.5833,
            1.4146,
            0.8986,
            0.5659,
            0.7069,
            0.5338,
            0.4889,
            0.4917,
            0.4069,
            0.4999,
            0.6866,
            0.4093,
            0.5709,
            0.6065,
            0.6415,
            0.4944,
            0.5726,
            1.2042,
            0.5458,
            1.6887,
            0.3971,
            1.0600,
            0.3943,
            0.5537,
            0.5444,
            0.4089,
            0.7468,
            0.7744,
        ]

        self.mean = torch.tensor(mean, dtype=dtype, device=device)
        self.std = torch.tensor(std, dtype=dtype, device=device)
        self.scale = [self.mean, 1.0 / self.std]

        # init model
        self.model = _video_vae(
            pretrained_path=vae_pth,
            s3_credential_path=s3_credential_path,
            device=device,
            temporal_window=temporal_window,
        )
        self.model = self.model.eval().requires_grad_(False)
        self.is_amp = is_amp
        if not is_amp:
            self.model = self.model.to(dtype=dtype)
            self.context = nullcontext()
        else:
            self.context = torch.amp.autocast("cuda", dtype=dtype)

    def count_param(self):
        return sum(p.numel() for p in self.model.parameters())

    @torch.no_grad()
    def encode(self, videos):
        """
        videos: A list of videos each with shape [C, T, H, W].
        """
        if self.benchmark:
            torch.cuda.synchronize()
            benchmark_times = BenchmarkTimes()
            total_time = time.perf_counter()
        in_dtype = videos.dtype
        with self.context:
            if not self.is_amp:
                videos = videos.to(self.dtype)
            if self.benchmark:
                torch.cuda.synchronize()
                model_time = time.perf_counter()
            latent = self.model.encode(videos, self.scale)
            if self.benchmark:
                torch.cuda.synchronize()
                benchmark_times.model_invocation = time.perf_counter() - model_time
        latent = latent.to(in_dtype)
        if self.benchmark:
            torch.cuda.synchronize()
            benchmark_times.total = time.perf_counter() - total_time
            return latent, benchmark_times
        return latent

    @torch.no_grad()
    def decode(self, zs):
        if self.benchmark:
            torch.cuda.synchronize()
            benchmark_times = BenchmarkTimes()
            total_time = time.perf_counter()
        in_dtype = zs.dtype
        with self.context:
            if not self.is_amp:
                zs = zs.to(self.dtype)
            if self.benchmark:
                torch.cuda.synchronize()
                model_time = time.perf_counter()
            video_recon = self.model.decode(zs, self.scale)
            if self.benchmark:
                torch.cuda.synchronize()
                benchmark_times.model_invocation = time.perf_counter() - model_time
        video_recon = video_recon.to(in_dtype)
        if self.benchmark:
            torch.cuda.synchronize()
            benchmark_times.total = time.perf_counter() - total_time
            return video_recon, benchmark_times
        return video_recon


class Wan2pt2VAEInterface(VideoTokenizerInterface):
    def __init__(self, chunk_duration: int = 93, **kwargs):
        self.model = WanVAE(
            dtype=torch.bfloat16,
            is_amp=False,
            vae_pth=kwargs.get(
                "vae_pth",
                "s3://bucket/cosmos_diffusion_v2/pretrain_weights/tokenizer/wan2pt2/Wan2.2_VAE.pth",
            ),
            s3_credential_path=kwargs.get("s3_credential_path", "credentials/s3_training.secret"),
            temporal_window=kwargs.get("temporal_window", 4),
        )

        del kwargs
        self.chunk_duration = chunk_duration

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

    def reset_dtype(self):
        pass

    def encode(self, state: torch.Tensor) -> torch.Tensor:
        latents = self.model.encode(state)
        return latents

    def decode(self, latent: torch.Tensor) -> torch.Tensor:
        return self.model.decode(latent)

    def get_latent_num_frames(self, num_pixel_frames: int) -> int:
        return 1 + (num_pixel_frames - 1) // 4

    def get_pixel_num_frames(self, num_latent_frames: int) -> int:
        return (num_latent_frames - 1) * 4 + 1

    @property
    def spatial_compression_factor(self):
        return 16  # 2x from patchify + 8x from spatial downsampling

    @property
    def temporal_compression_factor(self):
        return 4

    @property
    def pixel_chunk_duration(self):
        return self.chunk_duration

    @property
    def latent_chunk_duration(self):
        return self.get_latent_num_frames(self.chunk_duration)

    @property
    def latent_ch(self):
        return 48

    @property
    def spatial_resolution(self):
        return 512

    @property
    def name(self):
        return "wan2pt2_tokenizer"