Upload 5 files

swin2_mose/model.py

@@ -0,0 +1,1157 @@
+#
+# Source code: https://github.com/mv-lab/swin2sr
+#
+# -----------------------------------------------------------------------------------
+# Swin2SR: Swin2SR: SwinV2 Transformer for Compressed Image Super-Resolution and Restoration, https://arxiv.org/abs/2209.11345
+# Written by Conde and Choi et al.
+# -----------------------------------------------------------------------------------
+
+import math
+import numpy as np
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+import torch.utils.checkpoint as checkpoint
+from timm.models.layers import DropPath, to_2tuple, trunc_normal_
+
+from utils import window_reverse, Mlp, window_partition
+from moe import MoE
+
+
+class WindowAttention(nn.Module):
+    r""" Window based multi-head self attention (W-MSA) module with relative position bias.
+    It supports both of shifted and non-shifted window.
+    Args:
+        dim (int): Number of input channels.
+        window_size (tuple[int]): The height and width of the window.
+        num_heads (int): Number of attention heads.
+        qkv_bias (bool, optional):  If True, add a learnable bias to query, key, value. Default: True
+        attn_drop (float, optional): Dropout ratio of attention weight. Default: 0.0
+        proj_drop (float, optional): Dropout ratio of output. Default: 0.0
+        pretrained_window_size (tuple[int]): The height and width of the window in pre-training.
+    """
+
+    def __init__(self, dim, window_size, num_heads, qkv_bias=True, attn_drop=0., proj_drop=0.,
+                 pretrained_window_size=[0, 0],
+                 use_lepe=False,
+                 use_cpb_bias=True,
+                 use_rpe_bias=False):
+
+        super().__init__()
+        self.dim = dim
+        self.window_size = window_size  # Wh, Ww
+        self.pretrained_window_size = pretrained_window_size
+        self.num_heads = num_heads
+
+        self.logit_scale = nn.Parameter(torch.log(10 * torch.ones((num_heads, 1, 1))), requires_grad=True)
+
+        self.use_cpb_bias = use_cpb_bias
+
+        if self.use_cpb_bias:
+            print('positional encoder: CPB')
+            # mlp to generate continuous relative position bias
+            self.cpb_mlp = nn.Sequential(nn.Linear(2, 512, bias=True),
+                                         nn.ReLU(inplace=True),
+                                         nn.Linear(512, num_heads, bias=False))
+
+            # get relative_coords_table
+            relative_coords_h = torch.arange(-(self.window_size[0] - 1), self.window_size[0], dtype=torch.float32)
+            relative_coords_w = torch.arange(-(self.window_size[1] - 1), self.window_size[1], dtype=torch.float32)
+            relative_coords_table = torch.stack(
+                torch.meshgrid([relative_coords_h,
+                                relative_coords_w])).permute(1, 2, 0).contiguous().unsqueeze(0)  # 1, 2*Wh-1, 2*Ww-1, 2
+            if pretrained_window_size[0] > 0:
+                relative_coords_table[:, :, :, 0] /= (pretrained_window_size[0] - 1)
+                relative_coords_table[:, :, :, 1] /= (pretrained_window_size[1] - 1)
+            else:
+                relative_coords_table[:, :, :, 0] /= (self.window_size[0] - 1)
+                relative_coords_table[:, :, :, 1] /= (self.window_size[1] - 1)
+            relative_coords_table *= 8  # normalize to -8, 8
+            relative_coords_table = torch.sign(relative_coords_table) * torch.log2(
+                torch.abs(relative_coords_table) + 1.0) / np.log2(8)
+
+            self.register_buffer("relative_coords_table", relative_coords_table)
+
+            # get pair-wise relative position index for each token inside the window
+            coords_h = torch.arange(self.window_size[0])
+            coords_w = torch.arange(self.window_size[1])
+            coords = torch.stack(torch.meshgrid([coords_h, coords_w]))  # 2, Wh, Ww
+            coords_flatten = torch.flatten(coords, 1)  # 2, Wh*Ww
+            relative_coords = coords_flatten[:, :, None] - coords_flatten[:, None, :]  # 2, Wh*Ww, Wh*Ww
+            relative_coords = relative_coords.permute(1, 2, 0).contiguous()  # Wh*Ww, Wh*Ww, 2
+            relative_coords[:, :, 0] += self.window_size[0] - 1  # shift to start from 0
+            relative_coords[:, :, 1] += self.window_size[1] - 1
+            relative_coords[:, :, 0] *= 2 * self.window_size[1] - 1
+            relative_position_index = relative_coords.sum(-1)  # Wh*Ww, Wh*Ww
+            self.register_buffer("relative_position_index", relative_position_index)
+
+        self.use_rpe_bias = use_rpe_bias
+        if self.use_rpe_bias:
+            print('positional encoder: RPE')
+            # define a parameter table of relative position bias
+            self.relative_position_bias_table = nn.Parameter(
+                torch.zeros((2 * window_size[0] - 1) * (2 * window_size[1] - 1), num_heads))  # 2*Wh-1 * 2*Ww-1, nH
+
+            # get pair-wise relative position index for each token inside the window
+            coords_h = torch.arange(self.window_size[0])
+            coords_w = torch.arange(self.window_size[1])
+            coords = torch.stack(torch.meshgrid([coords_h, coords_w]))  # 2, Wh, Ww
+            coords_flatten = torch.flatten(coords, 1)  # 2, Wh*Ww
+            relative_coords = coords_flatten[:, :, None] - coords_flatten[:, None, :]  # 2, Wh*Ww, Wh*Ww
+            relative_coords = relative_coords.permute(1, 2, 0).contiguous()  # Wh*Ww, Wh*Ww, 2
+            relative_coords[:, :, 0] += self.window_size[0] - 1  # shift to start from 0
+            relative_coords[:, :, 1] += self.window_size[1] - 1
+            relative_coords[:, :, 0] *= 2 * self.window_size[1] - 1
+            rpe_relative_position_index = relative_coords.sum(-1)  # Wh*Ww, Wh*Ww
+            self.register_buffer("rpe_relative_position_index", rpe_relative_position_index)
+
+            trunc_normal_(self.relative_position_bias_table, std=.02)
+
+        self.qkv = nn.Linear(dim, dim * 3, bias=False)
+        if qkv_bias:
+            self.q_bias = nn.Parameter(torch.zeros(dim))
+            self.v_bias = nn.Parameter(torch.zeros(dim))
+        else:
+            self.q_bias = None
+            self.v_bias = None
+        self.attn_drop = nn.Dropout(attn_drop)
+        self.proj = nn.Linear(dim, dim)
+        self.proj_drop = nn.Dropout(proj_drop)
+        self.softmax = nn.Softmax(dim=-1)
+
+        self.use_lepe = use_lepe
+        if self.use_lepe:
+            print('positional encoder: LEPE')
+            self.get_v = nn.Conv2d(
+                dim, dim, kernel_size=3, stride=1, padding=1, groups=dim)
+
+    def forward(self, x, mask=None):
+        """
+        Args:
+            x: input features with shape of (num_windows*B, N, C)
+            mask: (0/-inf) mask with shape of (num_windows, Wh*Ww, Wh*Ww) or None
+        """
+        B_, N, C = x.shape
+        qkv_bias = None
+        if self.q_bias is not None:
+            qkv_bias = torch.cat((self.q_bias, torch.zeros_like(self.v_bias, requires_grad=False), self.v_bias))
+        qkv = F.linear(input=x, weight=self.qkv.weight, bias=qkv_bias)
+        qkv = qkv.reshape(B_, N, 3, self.num_heads, -1).permute(2, 0, 3, 1, 4)
+        q, k, v = qkv[0], qkv[1], qkv[2]  # make torchscript happy (cannot use tensor as tuple)
+
+        if self.use_lepe:
+            lepe = self.lepe_pos(v)
+
+        # cosine attention
+        attn = (F.normalize(q, dim=-1) @ F.normalize(k, dim=-1).transpose(-2, -1))
+        logit_scale = torch.clamp(self.logit_scale, max=torch.log(torch.tensor(1. / 0.01)).to(self.logit_scale.device)).exp()
+        attn = attn * logit_scale
+
+        if self.use_cpb_bias:
+            relative_position_bias_table = self.cpb_mlp(self.relative_coords_table).view(-1, self.num_heads)
+            relative_position_bias = relative_position_bias_table[self.relative_position_index.view(-1)].view(
+                self.window_size[0] * self.window_size[1], self.window_size[0] * self.window_size[1], -1)  # Wh*Ww,Wh*Ww,nH
+            relative_position_bias = relative_position_bias.permute(2, 0, 1).contiguous()  # nH, Wh*Ww, Wh*Ww
+            relative_position_bias = 16 * torch.sigmoid(relative_position_bias)
+            attn = attn + relative_position_bias.unsqueeze(0)
+
+        if self.use_rpe_bias:
+            relative_position_bias = self.relative_position_bias_table[self.rpe_relative_position_index.view(-1)].view(
+                self.window_size[0] * self.window_size[1], self.window_size[0] * self.window_size[1], -1)  # Wh*Ww,Wh*Ww,nH
+            relative_position_bias = relative_position_bias.permute(2, 0, 1).contiguous()  # nH, Wh*Ww, Wh*Ww
+            attn = attn + relative_position_bias.unsqueeze(0)
+
+        if mask is not None:
+            nW = mask.shape[0]
+            attn = attn.view(B_ // nW, nW, self.num_heads, N, N) + mask.unsqueeze(1).unsqueeze(0)
+            attn = attn.view(-1, self.num_heads, N, N)
+            attn = self.softmax(attn)
+        else:
+            attn = self.softmax(attn)
+
+        attn = self.attn_drop(attn)
+
+        x = (attn @ v)
+
+        if self.use_lepe:
+            x = x + lepe
+
+        x = x.transpose(1, 2).reshape(B_, N, C)
+        x = self.proj(x)
+        x = self.proj_drop(x)
+        return x
+
+    def lepe_pos(self, v):
+        B, NH, HW, NW = v.shape
+        C = NH * NW
+        H = W = int(math.sqrt(HW))
+        v = v.transpose(-2, -1).contiguous().view(B, C, H, W)
+        lepe = self.get_v(v)
+        lepe = lepe.reshape(-1, self.num_heads, NW, HW)
+        lepe = lepe.permute(0, 1, 3, 2).contiguous()
+        return lepe
+
+    def extra_repr(self) -> str:
+        return f'dim={self.dim}, window_size={self.window_size}, ' \
+               f'pretrained_window_size={self.pretrained_window_size}, num_heads={self.num_heads}'
+
+    def flops(self, N):
+        # calculate flops for 1 window with token length of N
+        flops = 0
+        # qkv = self.qkv(x)
+        flops += N * self.dim * 3 * self.dim
+        # attn = (q @ k.transpose(-2, -1))
+        flops += self.num_heads * N * (self.dim // self.num_heads) * N
+        #  x = (attn @ v)
+        flops += self.num_heads * N * N * (self.dim // self.num_heads)
+        # x = self.proj(x)
+        flops += N * self.dim * self.dim
+        return flops
+
+
+class SwinTransformerBlock(nn.Module):
+    r""" Swin Transformer Block.
+    Args:
+        dim (int): Number of input channels.
+        input_resolution (tuple[int]): Input resulotion.
+        num_heads (int): Number of attention heads.
+        window_size (int): Window size.
+        shift_size (int): Shift size for SW-MSA.
+        mlp_ratio (float): Ratio of mlp hidden dim to embedding dim.
+        qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True
+        drop (float, optional): Dropout rate. Default: 0.0
+        attn_drop (float, optional): Attention dropout rate. Default: 0.0
+        drop_path (float, optional): Stochastic depth rate. Default: 0.0
+        act_layer (nn.Module, optional): Activation layer. Default: nn.GELU
+        norm_layer (nn.Module, optional): Normalization layer.  Default: nn.LayerNorm
+        pretrained_window_size (int): Window size in pre-training.
+    """
+
+    def __init__(self, dim, input_resolution, num_heads, window_size=7, shift_size=0,
+                 mlp_ratio=4., qkv_bias=True, drop=0., attn_drop=0., drop_path=0.,
+                 act_layer=nn.GELU, norm_layer=nn.LayerNorm, pretrained_window_size=0,
+                 use_lepe=False,
+                 use_cpb_bias=True,
+                 MoE_config=None,
+                 use_rpe_bias=False):
+        super().__init__()
+        self.dim = dim
+        self.input_resolution = input_resolution
+        self.num_heads = num_heads
+        self.window_size = window_size
+        self.shift_size = shift_size
+        self.mlp_ratio = mlp_ratio
+        if min(self.input_resolution) <= self.window_size:
+            # if window size is larger than input resolution, we don't partition windows
+            self.shift_size = 0
+            self.window_size = min(self.input_resolution)
+        assert 0 <= self.shift_size < self.window_size, "shift_size must in 0-window_size"
+
+        self.norm1 = norm_layer(dim)
+        self.attn = WindowAttention(
+            dim, window_size=to_2tuple(self.window_size), num_heads=num_heads,
+            qkv_bias=qkv_bias, attn_drop=attn_drop, proj_drop=drop,
+            pretrained_window_size=to_2tuple(pretrained_window_size),
+            use_lepe=use_lepe,
+            use_cpb_bias=use_cpb_bias,
+            use_rpe_bias=use_rpe_bias)
+
+        self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity()
+        self.norm2 = norm_layer(dim)
+        mlp_hidden_dim = int(dim * mlp_ratio)
+
+        if MoE_config is None:
+            print('-->>> MLP')
+            self.mlp = Mlp(
+                in_features=dim, hidden_features=mlp_hidden_dim,
+                act_layer=act_layer, drop=drop)
+        else:
+            print('-->>> MOE')
+            print(MoE_config)
+            self.mlp = MoE(
+                input_size=dim, output_size=dim, hidden_size=mlp_hidden_dim,
+                **MoE_config)
+
+        if self.shift_size > 0:
+            attn_mask = self.calculate_mask(self.input_resolution)
+        else:
+            attn_mask = None
+
+        self.register_buffer("attn_mask", attn_mask)
+
+    def calculate_mask(self, x_size):
+        # calculate attention mask for SW-MSA
+        H, W = x_size
+        img_mask = torch.zeros((1, H, W, 1))  # 1 H W 1
+        h_slices = (slice(0, -self.window_size),
+                    slice(-self.window_size, -self.shift_size),
+                    slice(-self.shift_size, None))
+        w_slices = (slice(0, -self.window_size),
+                    slice(-self.window_size, -self.shift_size),
+                    slice(-self.shift_size, None))
+        cnt = 0
+        for h in h_slices:
+            for w in w_slices:
+                img_mask[:, h, w, :] = cnt
+                cnt += 1
+
+        mask_windows = window_partition(img_mask, self.window_size)  # nW, window_size, window_size, 1
+        mask_windows = mask_windows.view(-1, self.window_size * self.window_size)
+        attn_mask = mask_windows.unsqueeze(1) - mask_windows.unsqueeze(2)
+        attn_mask = attn_mask.masked_fill(attn_mask != 0, float(-100.0)).masked_fill(attn_mask == 0, float(0.0))
+
+        return attn_mask
+
+    def forward(self, x, x_size):
+        H, W = x_size
+        B, L, C = x.shape
+
+        shortcut = x
+        x = x.view(B, H, W, C)
+
+        # cyclic shift
+        if self.shift_size > 0:
+            shifted_x = torch.roll(x, shifts=(-self.shift_size, -self.shift_size), dims=(1, 2))
+        else:
+            shifted_x = x
+
+        # partition windows
+        x_windows = window_partition(shifted_x, self.window_size)  # nW*B, window_size, window_size, C
+        x_windows = x_windows.view(-1, self.window_size * self.window_size, C)  # nW*B, window_size*window_size, C
+
+        # W-MSA/SW-MSA (to be compatible for testing on images whose shapes are the multiple of window size
+        if self.input_resolution == x_size:
+            attn_windows = self.attn(x_windows, mask=self.attn_mask)  # nW*B, window_size*window_size, C
+        else:
+            attn_windows = self.attn(x_windows, mask=self.calculate_mask(x_size).to(x.device))
+
+        # merge windows
+        attn_windows = attn_windows.view(-1, self.window_size, self.window_size, C)
+        shifted_x = window_reverse(attn_windows, self.window_size, H, W)  # B H' W' C
+
+        # reverse cyclic shift
+        if self.shift_size > 0:
+            x = torch.roll(shifted_x, shifts=(self.shift_size, self.shift_size), dims=(1, 2))
+        else:
+            x = shifted_x
+        x = x.view(B, H * W, C)
+        x = shortcut + self.drop_path(self.norm1(x))
+
+        # FFN
+
+        loss_moe = None
+        res = self.mlp(x)
+        if not torch.is_tensor(res):
+            res, loss_moe = res
+
+        x = x + self.drop_path(self.norm2(res))
+
+        return x, loss_moe
+
+    def extra_repr(self) -> str:
+        return f"dim={self.dim}, input_resolution={self.input_resolution}, num_heads={self.num_heads}, " \
+               f"window_size={self.window_size}, shift_size={self.shift_size}, mlp_ratio={self.mlp_ratio}"
+
+    def flops(self):
+        flops = 0
+        H, W = self.input_resolution
+        # norm1
+        flops += self.dim * H * W
+        # W-MSA/SW-MSA
+        nW = H * W / self.window_size / self.window_size
+        flops += nW * self.attn.flops(self.window_size * self.window_size)
+        # mlp
+        flops += 2 * H * W * self.dim * self.dim * self.mlp_ratio
+        # norm2
+        flops += self.dim * H * W
+        return flops
+
+
+class PatchMerging(nn.Module):
+    r""" Patch Merging Layer.
+    Args:
+        input_resolution (tuple[int]): Resolution of input feature.
+        dim (int): Number of input channels.
+        norm_layer (nn.Module, optional): Normalization layer.  Default: nn.LayerNorm
+    """
+
+    def __init__(self, input_resolution, dim, norm_layer=nn.LayerNorm):
+        super().__init__()
+        self.input_resolution = input_resolution
+        self.dim = dim
+        self.reduction = nn.Linear(4 * dim, 2 * dim, bias=False)
+        self.norm = norm_layer(2 * dim)
+
+    def forward(self, x):
+        """
+        x: B, H*W, C
+        """
+        H, W = self.input_resolution
+        B, L, C = x.shape
+        assert L == H * W, "input feature has wrong size"
+        assert H % 2 == 0 and W % 2 == 0, f"x size ({H}*{W}) are not even."
+
+        x = x.view(B, H, W, C)
+
+        x0 = x[:, 0::2, 0::2, :]  # B H/2 W/2 C
+        x1 = x[:, 1::2, 0::2, :]  # B H/2 W/2 C
+        x2 = x[:, 0::2, 1::2, :]  # B H/2 W/2 C
+        x3 = x[:, 1::2, 1::2, :]  # B H/2 W/2 C
+        x = torch.cat([x0, x1, x2, x3], -1)  # B H/2 W/2 4*C
+        x = x.view(B, -1, 4 * C)  # B H/2*W/2 4*C
+
+        x = self.reduction(x)
+        x = self.norm(x)
+
+        return x
+
+    def extra_repr(self) -> str:
+        return f"input_resolution={self.input_resolution}, dim={self.dim}"
+
+    def flops(self):
+        H, W = self.input_resolution
+        flops = (H // 2) * (W // 2) * 4 * self.dim * 2 * self.dim
+        flops += H * W * self.dim // 2
+        return flops
+
+
+class BasicLayer(nn.Module):
+    """ A basic Swin Transformer layer for one stage.
+    Args:
+        dim (int): Number of input channels.
+        input_resolution (tuple[int]): Input resolution.
+        depth (int): Number of blocks.
+        num_heads (int): Number of attention heads.
+        window_size (int): Local window size.
+        mlp_ratio (float): Ratio of mlp hidden dim to embedding dim.
+        qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True
+        drop (float, optional): Dropout rate. Default: 0.0
+        attn_drop (float, optional): Attention dropout rate. Default: 0.0
+        drop_path (float | tuple[float], optional): Stochastic depth rate. Default: 0.0
+        norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm
+        downsample (nn.Module | None, optional): Downsample layer at the end of the layer. Default: None
+        use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False.
+        pretrained_window_size (int): Local window size in pre-training.
+    """
+
+    def __init__(self, dim, input_resolution, depth, num_heads, window_size,
+                 mlp_ratio=4., qkv_bias=True, drop=0., attn_drop=0.,
+                 drop_path=0., norm_layer=nn.LayerNorm, downsample=None, use_checkpoint=False,
+                 pretrained_window_size=0,
+                 use_lepe=False,
+                 use_cpb_bias=True,
+                 MoE_config=None,
+                 use_rpe_bias=False):
+
+        super().__init__()
+        self.dim = dim
+        self.input_resolution = input_resolution
+        self.depth = depth
+        self.use_checkpoint = use_checkpoint
+
+        # build blocks
+        self.blocks = nn.ModuleList([
+            SwinTransformerBlock(dim=dim, input_resolution=input_resolution,
+                                 num_heads=num_heads, window_size=window_size,
+                                 shift_size=0 if (i % 2 == 0) else window_size // 2,
+                                 mlp_ratio=mlp_ratio,
+                                 qkv_bias=qkv_bias,
+                                 drop=drop, attn_drop=attn_drop,
+                                 drop_path=drop_path[i] if isinstance(drop_path, list) else drop_path,
+                                 norm_layer=norm_layer,
+                                 pretrained_window_size=pretrained_window_size,
+                                 use_lepe=use_lepe,
+                                 use_cpb_bias=use_cpb_bias,
+                                 MoE_config=MoE_config,
+                                 use_rpe_bias=use_rpe_bias)
+            for i in range(depth)])
+
+        # patch merging layer
+        if downsample is not None:
+            self.downsample = downsample(input_resolution, dim=dim, norm_layer=norm_layer)
+        else:
+            self.downsample = None
+
+    def forward(self, x, x_size):
+        loss_moe_all = 0
+        for blk in self.blocks:
+            if self.use_checkpoint:
+                x = checkpoint.checkpoint(blk, x, x_size)
+            else:
+                x = blk(x, x_size)
+
+            if not torch.is_tensor(x):
+                x, loss_moe = x
+                loss_moe_all += loss_moe or 0
+
+        if self.downsample is not None:
+            x = self.downsample(x)
+        return x, loss_moe_all
+
+    def extra_repr(self) -> str:
+        return f"dim={self.dim}, input_resolution={self.input_resolution}, depth={self.depth}"
+
+    def flops(self):
+        flops = 0
+        for blk in self.blocks:
+            flops += blk.flops()
+        if self.downsample is not None:
+            flops += self.downsample.flops()
+        return flops
+
+    def _init_respostnorm(self):
+        for blk in self.blocks:
+            nn.init.constant_(blk.norm1.bias, 0)
+            nn.init.constant_(blk.norm1.weight, 0)
+            nn.init.constant_(blk.norm2.bias, 0)
+            nn.init.constant_(blk.norm2.weight, 0)
+
+class PatchEmbed(nn.Module):
+    r""" Image to Patch Embedding
+    Args:
+        img_size (int): Image size.  Default: 224.
+        patch_size (int): Patch token size. Default: 4.
+        in_chans (int): Number of input image channels. Default: 3.
+        embed_dim (int): Number of linear projection output channels. Default: 96.
+        norm_layer (nn.Module, optional): Normalization layer. Default: None
+    """
+
+    def __init__(self, img_size=224, patch_size=4, in_chans=3, embed_dim=96, norm_layer=None):
+        super().__init__()
+        img_size = to_2tuple(img_size)
+        patch_size = to_2tuple(patch_size)
+        patches_resolution = [img_size[0] // patch_size[0], img_size[1] // patch_size[1]]
+        self.img_size = img_size
+        self.patch_size = patch_size
+        self.patches_resolution = patches_resolution
+        self.num_patches = patches_resolution[0] * patches_resolution[1]
+
+        self.in_chans = in_chans
+        self.embed_dim = embed_dim
+
+        self.proj = nn.Conv2d(in_chans, embed_dim, kernel_size=patch_size, stride=patch_size)
+        if norm_layer is not None:
+            self.norm = norm_layer(embed_dim)
+        else:
+            self.norm = None
+
+    def forward(self, x):
+        B, C, H, W = x.shape
+        # FIXME look at relaxing size constraints
+        # assert H == self.img_size[0] and W == self.img_size[1],
+        #     f"Input image size ({H}*{W}) doesn't match model ({self.img_size[0]}*{self.img_size[1]})."
+        x = self.proj(x).flatten(2).transpose(1, 2)  # B Ph*Pw C
+        if self.norm is not None:
+            x = self.norm(x)
+        return x
+
+    def flops(self):
+        Ho, Wo = self.patches_resolution
+        flops = Ho * Wo * self.embed_dim * self.in_chans * (self.patch_size[0] * self.patch_size[1])
+        if self.norm is not None:
+            flops += Ho * Wo * self.embed_dim
+        return flops
+
+
+class RSTB(nn.Module):
+    """Residual Swin Transformer Block (RSTB).
+
+    Args:
+        dim (int): Number of input channels.
+        input_resolution (tuple[int]): Input resolution.
+        depth (int): Number of blocks.
+        num_heads (int): Number of attention heads.
+        window_size (int): Local window size.
+        mlp_ratio (float): Ratio of mlp hidden dim to embedding dim.
+        qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True
+        drop (float, optional): Dropout rate. Default: 0.0
+        attn_drop (float, optional): Attention dropout rate. Default: 0.0
+        drop_path (float | tuple[float], optional): Stochastic depth rate. Default: 0.0
+        norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm
+        downsample (nn.Module | None, optional): Downsample layer at the end of the layer. Default: None
+        use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False.
+        img_size: Input image size.
+        patch_size: Patch size.
+        resi_connection: The convolutional block before residual connection.
+    """
+
+    def __init__(self, dim, input_resolution, depth, num_heads, window_size,
+                 mlp_ratio=4., qkv_bias=True, drop=0., attn_drop=0.,
+                 drop_path=0., norm_layer=nn.LayerNorm, downsample=None, use_checkpoint=False,
+                 img_size=224, patch_size=4, resi_connection='1conv',
+                 use_lepe=False,
+                 use_cpb_bias=True,
+                 MoE_config=None,
+                 use_rpe_bias=False):
+        super(RSTB, self).__init__()
+
+        self.dim = dim
+        self.input_resolution = input_resolution
+
+        self.residual_group = BasicLayer(dim=dim,
+                                         input_resolution=input_resolution,
+                                         depth=depth,
+                                         num_heads=num_heads,
+                                         window_size=window_size,
+                                         mlp_ratio=mlp_ratio,
+                                         qkv_bias=qkv_bias,
+                                         drop=drop, attn_drop=attn_drop,
+                                         drop_path=drop_path,
+                                         norm_layer=norm_layer,
+                                         downsample=downsample,
+                                         use_checkpoint=use_checkpoint,
+                                         use_lepe=use_lepe,
+                                         use_cpb_bias=use_cpb_bias,
+                                         MoE_config=MoE_config,
+                                         use_rpe_bias=use_rpe_bias
+                                         )
+
+        if resi_connection == '1conv':
+            self.conv = nn.Conv2d(dim, dim, 3, 1, 1)
+        elif resi_connection == '3conv':
+            # to save parameters and memory
+            self.conv = nn.Sequential(nn.Conv2d(dim, dim // 4, 3, 1, 1), nn.LeakyReLU(negative_slope=0.2, inplace=True),
+                                      nn.Conv2d(dim // 4, dim // 4, 1, 1, 0),
+                                      nn.LeakyReLU(negative_slope=0.2, inplace=True),
+                                      nn.Conv2d(dim // 4, dim, 3, 1, 1))
+
+        self.patch_embed = PatchEmbed(
+            img_size=img_size, patch_size=patch_size, in_chans=dim, embed_dim=dim,
+            norm_layer=None)
+
+        self.patch_unembed = PatchUnEmbed(
+            img_size=img_size, patch_size=patch_size, in_chans=dim, embed_dim=dim,
+            norm_layer=None)
+
+    def forward(self, x, x_size):
+        loss_moe = None
+        res = self.residual_group(x, x_size)
+
+        if not torch.is_tensor(res):
+            res, loss_moe = res
+
+        res = self.patch_embed(self.conv(self.patch_unembed(res, x_size)))
+        return res + x, loss_moe
+
+    def flops(self):
+        flops = 0
+        flops += self.residual_group.flops()
+        H, W = self.input_resolution
+        flops += H * W * self.dim * self.dim * 9
+        flops += self.patch_embed.flops()
+        flops += self.patch_unembed.flops()
+
+        return flops
+
+
+class PatchUnEmbed(nn.Module):
+    r""" Image to Patch Unembedding
+
+    Args:
+        img_size (int): Image size.  Default: 224.
+        patch_size (int): Patch token size. Default: 4.
+        in_chans (int): Number of input image channels. Default: 3.
+        embed_dim (int): Number of linear projection output channels. Default: 96.
+        norm_layer (nn.Module, optional): Normalization layer. Default: None
+    """
+
+    def __init__(self, img_size=224, patch_size=4, in_chans=3, embed_dim=96, norm_layer=None):
+        super().__init__()
+        img_size = to_2tuple(img_size)
+        patch_size = to_2tuple(patch_size)
+        patches_resolution = [img_size[0] // patch_size[0], img_size[1] // patch_size[1]]
+        self.img_size = img_size
+        self.patch_size = patch_size
+        self.patches_resolution = patches_resolution
+        self.num_patches = patches_resolution[0] * patches_resolution[1]
+
+        self.in_chans = in_chans
+        self.embed_dim = embed_dim
+
+    def forward(self, x, x_size):
+        B, HW, C = x.shape
+        x = x.transpose(1, 2).view(B, self.embed_dim, x_size[0], x_size[1])  # B Ph*Pw C
+        return x
+
+    def flops(self):
+        flops = 0
+        return flops
+
+
+class Upsample(nn.Sequential):
+    """Upsample module.
+
+    Args:
+        scale (int): Scale factor. Supported scales: 2^n and 3.
+        num_feat (int): Channel number of intermediate features.
+    """
+
+    def __init__(self, scale, num_feat):
+        m = []
+        if (scale & (scale - 1)) == 0:  # scale = 2^n
+            for _ in range(int(math.log(scale, 2))):
+                m.append(nn.Conv2d(num_feat, 4 * num_feat, 3, 1, 1))
+                m.append(nn.PixelShuffle(2))
+        elif scale == 3:
+            m.append(nn.Conv2d(num_feat, 9 * num_feat, 3, 1, 1))
+            m.append(nn.PixelShuffle(3))
+        else:
+            raise ValueError(f'scale {scale} is not supported. ' 'Supported scales: 2^n and 3.')
+        super(Upsample, self).__init__(*m)
+
+class Upsample_hf(nn.Sequential):
+    """Upsample module.
+
+    Args:
+        scale (int): Scale factor. Supported scales: 2^n and 3.
+        num_feat (int): Channel number of intermediate features.
+    """
+
+    def __init__(self, scale, num_feat):
+        m = []
+        if (scale & (scale - 1)) == 0:  # scale = 2^n
+            for _ in range(int(math.log(scale, 2))):
+                m.append(nn.Conv2d(num_feat, 4 * num_feat, 3, 1, 1))
+                m.append(nn.PixelShuffle(2))
+        elif scale == 3:
+            m.append(nn.Conv2d(num_feat, 9 * num_feat, 3, 1, 1))
+            m.append(nn.PixelShuffle(3))
+        else:
+            raise ValueError(f'scale {scale} is not supported. ' 'Supported scales: 2^n and 3.')
+        super(Upsample_hf, self).__init__(*m)
+
+
+class UpsampleOneStep(nn.Sequential):
+    """UpsampleOneStep module (the difference with Upsample is that it always only has 1conv + 1pixelshuffle)
+       Used in lightweight SR to save parameters.
+
+    Args:
+        scale (int): Scale factor. Supported scales: 2^n and 3.
+        num_feat (int): Channel number of intermediate features.
+
+    """
+
+    def __init__(self, scale, num_feat, num_out_ch, input_resolution=None):
+        self.num_feat = num_feat
+        self.input_resolution = input_resolution
+        m = []
+        m.append(nn.Conv2d(num_feat, (scale ** 2) * num_out_ch, 3, 1, 1))
+        m.append(nn.PixelShuffle(scale))
+        super(UpsampleOneStep, self).__init__(*m)
+
+    def flops(self):
+        H, W = self.input_resolution
+        flops = H * W * self.num_feat * 3 * 9
+        return flops
+
+
+
+class Swin2SR(nn.Module):
+    r""" Swin2SR
+        A PyTorch impl of : `Swin2SR: SwinV2 Transformer for Compressed Image Super-Resolution and Restoration`.
+
+    Args:
+        img_size (int | tuple(int)): Input image size. Default 64
+        patch_size (int | tuple(int)): Patch size. Default: 1
+        in_chans (int): Number of input image channels. Default: 3
+        embed_dim (int): Patch embedding dimension. Default: 96
+        depths (tuple(int)): Depth of each Swin Transformer layer.
+        num_heads (tuple(int)): Number of attention heads in different layers.
+        window_size (int): Window size. Default: 7
+        mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. Default: 4
+        qkv_bias (bool): If True, add a learnable bias to query, key, value. Default: True
+        drop_rate (float): Dropout rate. Default: 0
+        attn_drop_rate (float): Attention dropout rate. Default: 0
+        drop_path_rate (float): Stochastic depth rate. Default: 0.1
+        norm_layer (nn.Module): Normalization layer. Default: nn.LayerNorm.
+        ape (bool): If True, add absolute position embedding to the patch embedding. Default: False
+        patch_norm (bool): If True, add normalization after patch embedding. Default: True
+        use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False
+        upscale: Upscale factor. 2/3/4/8 for image SR, 1 for denoising and compress artifact reduction
+        img_range: Image range. 1. or 255.
+        upsampler: The reconstruction reconstruction module. 'pixelshuffle'/'pixelshuffledirect'/'nearest+conv'/None
+        resi_connection: The convolutional block before residual connection. '1conv'/'3conv'
+    """
+
+    def __init__(self, img_size=64, patch_size=1, in_chans=3,
+                 embed_dim=96, depths=[6, 6, 6, 6], num_heads=[6, 6, 6, 6],
+                 window_size=7, mlp_ratio=4., qkv_bias=True,
+                 drop_rate=0., attn_drop_rate=0., drop_path_rate=0.1,
+                 norm_layer=nn.LayerNorm, ape=False, patch_norm=True,
+                 use_checkpoint=False, upscale=2, img_range=1., upsampler='', resi_connection='1conv',
+                 use_lepe=False,
+                 use_cpb_bias=True,
+                 MoE_config=None,
+                 use_rpe_bias=False,
+                 **kwargs):
+        super(Swin2SR, self).__init__()
+        print('==== SWIN 2SR')
+        num_in_ch = in_chans
+        num_out_ch = in_chans
+        num_feat = 64
+        self.img_range = img_range
+        if in_chans == 3:
+            rgb_mean = (0.4488, 0.4371, 0.4040)
+            self.mean = torch.Tensor(rgb_mean).view(1, 3, 1, 1)
+        else:
+            self.mean = torch.zeros(1, 1, 1, 1)
+        self.upscale = upscale
+        self.upsampler = upsampler
+        self.window_size = window_size
+
+        #####################################################################################################
+        ################################### 1, shallow feature extraction ###################################
+        self.conv_first = nn.Conv2d(num_in_ch, embed_dim, 3, 1, 1)
+
+        #####################################################################################################
+        ################################### 2, deep feature extraction ######################################
+        self.num_layers = len(depths)
+        self.embed_dim = embed_dim
+        self.ape = ape
+        self.patch_norm = patch_norm
+        self.num_features = embed_dim
+        self.mlp_ratio = mlp_ratio
+
+        # split image into non-overlapping patches
+        self.patch_embed = PatchEmbed(
+            img_size=img_size, patch_size=patch_size, in_chans=embed_dim, embed_dim=embed_dim,
+            norm_layer=norm_layer if self.patch_norm else None)
+        num_patches = self.patch_embed.num_patches
+        patches_resolution = self.patch_embed.patches_resolution
+        self.patches_resolution = patches_resolution
+
+        # merge non-overlapping patches into image
+        self.patch_unembed = PatchUnEmbed(
+            img_size=img_size, patch_size=patch_size, in_chans=embed_dim, embed_dim=embed_dim,
+            norm_layer=norm_layer if self.patch_norm else None)
+
+        # absolute position embedding
+        if self.ape:
+            self.absolute_pos_embed = nn.Parameter(torch.zeros(1, num_patches, embed_dim))
+            trunc_normal_(self.absolute_pos_embed, std=.02)
+
+        self.pos_drop = nn.Dropout(p=drop_rate)
+
+        # stochastic depth
+        dpr = [x.item() for x in torch.linspace(0, drop_path_rate, sum(depths))]  # stochastic depth decay rule
+
+        # build Residual Swin Transformer blocks (RSTB)
+        self.layers = nn.ModuleList()
+        for i_layer in range(self.num_layers):
+            layer = RSTB(dim=embed_dim,
+                         input_resolution=(patches_resolution[0],
+                                           patches_resolution[1]),
+                         depth=depths[i_layer],
+                         num_heads=num_heads[i_layer],
+                         window_size=window_size,
+                         mlp_ratio=self.mlp_ratio,
+                         qkv_bias=qkv_bias,
+                         drop=drop_rate, attn_drop=attn_drop_rate,
+                         drop_path=dpr[sum(depths[:i_layer]):sum(depths[:i_layer + 1])],  # no impact on SR results
+                         norm_layer=norm_layer,
+                         downsample=None,
+                         use_checkpoint=use_checkpoint,
+                         img_size=img_size,
+                         patch_size=patch_size,
+                         resi_connection=resi_connection,
+                         use_lepe=use_lepe,
+                         use_cpb_bias=use_cpb_bias,
+                         MoE_config=MoE_config,
+                         use_rpe_bias=use_rpe_bias,
+                         )
+            self.layers.append(layer)
+
+        if self.upsampler == 'pixelshuffle_hf':
+            self.layers_hf = nn.ModuleList()
+            for i_layer in range(self.num_layers):
+                layer = RSTB(dim=embed_dim,
+                             input_resolution=(patches_resolution[0],
+                                               patches_resolution[1]),
+                             depth=depths[i_layer],
+                             num_heads=num_heads[i_layer],
+                             window_size=window_size,
+                             mlp_ratio=self.mlp_ratio,
+                             qkv_bias=qkv_bias,
+                             drop=drop_rate, attn_drop=attn_drop_rate,
+                             drop_path=dpr[sum(depths[:i_layer]):sum(depths[:i_layer + 1])],  # no impact on SR results
+                             norm_layer=norm_layer,
+                             downsample=None,
+                             use_checkpoint=use_checkpoint,
+                             img_size=img_size,
+                             patch_size=patch_size,
+                             resi_connection=resi_connection,
+                             use_lepe=use_lepe,
+                             use_cpb_bias=use_cpb_bias,
+                             MoE_config=MoE_config,
+                             use_rpe_bias=use_rpe_bias
+                             )
+                self.layers_hf.append(layer)
+
+        self.norm = norm_layer(self.num_features)
+
+        # build the last conv layer in deep feature extraction
+        if resi_connection == '1conv':
+            self.conv_after_body = nn.Conv2d(embed_dim, embed_dim, 3, 1, 1)
+        elif resi_connection == '3conv':
+            # to save parameters and memory
+            self.conv_after_body = nn.Sequential(nn.Conv2d(embed_dim, embed_dim // 4, 3, 1, 1),
+                                                 nn.LeakyReLU(negative_slope=0.2, inplace=True),
+                                                 nn.Conv2d(embed_dim // 4, embed_dim // 4, 1, 1, 0),
+                                                 nn.LeakyReLU(negative_slope=0.2, inplace=True),
+                                                 nn.Conv2d(embed_dim // 4, embed_dim, 3, 1, 1))
+
+        #####################################################################################################
+        ################################ 3, high quality image reconstruction ################################
+        if self.upsampler == 'pixelshuffle':
+            # for classical SR
+            self.conv_before_upsample = nn.Sequential(nn.Conv2d(embed_dim, num_feat, 3, 1, 1),
+                                                      nn.LeakyReLU(inplace=True))
+            self.upsample = Upsample(upscale, num_feat)
+            self.conv_last = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1)
+        elif self.upsampler == 'pixelshuffle_aux':
+            self.conv_bicubic = nn.Conv2d(num_in_ch, num_feat, 3, 1, 1)
+            self.conv_before_upsample = nn.Sequential(
+                nn.Conv2d(embed_dim, num_feat, 3, 1, 1),
+                nn.LeakyReLU(inplace=True))
+            self.conv_aux = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1)
+            self.conv_after_aux = nn.Sequential(
+                nn.Conv2d(3, num_feat, 3, 1, 1),
+                nn.LeakyReLU(inplace=True))
+            self.upsample = Upsample(upscale, num_feat)
+            self.conv_last = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1)
+
+        elif self.upsampler == 'pixelshuffle_hf':
+            self.conv_before_upsample = nn.Sequential(nn.Conv2d(embed_dim, num_feat, 3, 1, 1),
+                                                      nn.LeakyReLU(inplace=True))
+            self.upsample = Upsample(upscale, num_feat)
+            self.upsample_hf = Upsample_hf(upscale, num_feat)
+            self.conv_last = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1)
+            self.conv_first_hf = nn.Sequential(nn.Conv2d(num_feat, embed_dim, 3, 1, 1),
+                                                      nn.LeakyReLU(inplace=True))
+            self.conv_after_body_hf = nn.Conv2d(embed_dim, embed_dim, 3, 1, 1)
+            self.conv_before_upsample_hf = nn.Sequential(
+                nn.Conv2d(embed_dim, num_feat, 3, 1, 1),
+                nn.LeakyReLU(inplace=True))
+            self.conv_last_hf = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1)
+
+        elif self.upsampler == 'pixelshuffledirect':
+            # for lightweight SR (to save parameters)
+            self.upsample = UpsampleOneStep(upscale, embed_dim, num_out_ch,
+                                            (patches_resolution[0], patches_resolution[1]))
+        elif self.upsampler == 'nearest+conv':
+            # for real-world SR (less artifacts)
+            assert self.upscale == 4, 'only support x4 now.'
+            self.conv_before_upsample = nn.Sequential(nn.Conv2d(embed_dim, num_feat, 3, 1, 1),
+                                                      nn.LeakyReLU(inplace=True))
+            self.conv_up1 = nn.Conv2d(num_feat, num_feat, 3, 1, 1)
+            self.conv_up2 = nn.Conv2d(num_feat, num_feat, 3, 1, 1)
+            self.conv_hr = nn.Conv2d(num_feat, num_feat, 3, 1, 1)
+            self.conv_last = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1)
+            self.lrelu = nn.LeakyReLU(negative_slope=0.2, inplace=True)
+        else:
+            # for image denoising and JPEG compression artifact reduction
+            self.conv_last = nn.Conv2d(embed_dim, num_out_ch, 3, 1, 1)
+
+        self.apply(self._init_weights)
+
+    def _init_weights(self, m):
+        if isinstance(m, nn.Linear):
+            trunc_normal_(m.weight, std=.02)
+            if isinstance(m, nn.Linear) and m.bias is not None:
+                nn.init.constant_(m.bias, 0)
+        elif isinstance(m, nn.LayerNorm):
+            nn.init.constant_(m.bias, 0)
+            nn.init.constant_(m.weight, 1.0)
+
+    @torch.jit.ignore
+    def no_weight_decay(self):
+        return {'absolute_pos_embed'}
+
+    @torch.jit.ignore
+    def no_weight_decay_keywords(self):
+        return {'relative_position_bias_table'}
+
+    def check_image_size(self, x):
+        _, _, h, w = x.size()
+        mod_pad_h = (self.window_size - h % self.window_size) % self.window_size
+        mod_pad_w = (self.window_size - w % self.window_size) % self.window_size
+        x = F.pad(x, (0, mod_pad_w, 0, mod_pad_h), 'reflect')
+        return x
+
+    def forward_features(self, x):
+        x_size = (x.shape[2], x.shape[3])
+        x = self.patch_embed(x)
+        if self.ape:
+            x = x + self.absolute_pos_embed
+        x = self.pos_drop(x)
+
+        loss_moe_all = 0
+        for layer in self.layers:
+            x = layer(x, x_size)
+
+            if not torch.is_tensor(x):
+                x, loss_moe = x
+                loss_moe_all += loss_moe or 0
+
+        x = self.norm(x)  # B L C
+        x = self.patch_unembed(x, x_size)
+
+        return x, loss_moe_all
+
+    def forward_features_hf(self, x):
+        x_size = (x.shape[2], x.shape[3])
+        x = self.patch_embed(x)
+        if self.ape:
+            x = x + self.absolute_pos_embed
+        x = self.pos_drop(x)
+
+        loss_moe_all = 0
+        for layer in self.layers_hf:
+            x = layer(x, x_size)
+
+            if not torch.is_tensor(x):
+                x, loss_moe = x
+                loss_moe_all += loss_moe or 0
+
+        x = self.norm(x)  # B L C
+        x = self.patch_unembed(x, x_size)
+
+        return x, loss_moe_all
+
+    def forward_backbone(self, x):
+        H, W = x.shape[2:]
+        x = self.check_image_size(x)
+
+        self.mean = self.mean.type_as(x)
+        x = (x - self.mean) * self.img_range
+
+        if self.upsampler == 'pixelshuffledirect':
+            # for lightweight SR
+            x = self.conv_first(x)
+
+            res = self.forward_features(x)
+            if not torch.is_tensor(res):
+                res, loss_moe = res
+
+            x = self.conv_after_body(res) + x
+        else:
+            raise Exception('not implemented yet')
+
+        x = x / self.img_range + self.mean
+        return x
+
+    def forward(self, x):
+        H, W = x.shape[2:]
+        x = self.check_image_size(x)
+
+        self.mean = self.mean.type_as(x)
+        x = (x - self.mean) * self.img_range
+
+        loss_moe = 0
+        if self.upsampler == 'pixelshuffle':
+            # for classical SR
+            x = self.conv_first(x)
+
+            res = self.forward_features(x)
+            if not torch.is_tensor(res):
+                res, loss_moe = res
+
+            x = self.conv_after_body(res) + x
+            x = self.conv_before_upsample(x)
+            x = self.conv_last(self.upsample(x))
+        elif self.upsampler == 'pixelshuffle_aux':
+            bicubic = F.interpolate(x, size=(H * self.upscale, W * self.upscale), mode='bicubic', align_corners=False)
+            bicubic = self.conv_bicubic(bicubic)
+            x = self.conv_first(x)
+
+            res = self.forward_features(x)
+            if not torch.is_tensor(res):
+                res, loss_moe = res
+
+            x = self.conv_after_body(res) + x
+            x = self.conv_before_upsample(x)
+            aux = self.conv_aux(x) # b, 3, LR_H, LR_W
+            x = self.conv_after_aux(aux)
+            x = self.upsample(x)[:, :, :H * self.upscale, :W * self.upscale] + bicubic[:, :, :H * self.upscale, :W * self.upscale]
+            x = self.conv_last(x)
+            aux = aux / self.img_range + self.mean
+        elif self.upsampler == 'pixelshuffle_hf':
+            # for classical SR with HF
+            x = self.conv_first(x)
+
+            res = self.forward_features(x)
+            if not torch.is_tensor(res):
+                res, loss_moe = res
+
+            x = self.conv_after_body(res) + x
+            x_before = self.conv_before_upsample(x)
+            x_out = self.conv_last(self.upsample(x_before))
+
+            x_hf = self.conv_first_hf(x_before)
+
+            res_hf = self.forward_features_hf(x_hf)
+            if not torch.is_tensor(res_hf):
+                res_hf, loss_moe_hf = res_hf
+                loss_moe += loss_moe_hf
+
+            x_hf = self.conv_after_body_hf(res_hf) + x_hf
+            x_hf = self.conv_before_upsample_hf(x_hf)
+            x_hf = self.conv_last_hf(self.upsample_hf(x_hf))
+            x = x_out + x_hf
+            x_hf = x_hf / self.img_range + self.mean
+
+        elif self.upsampler == 'pixelshuffledirect':
+            # for lightweight SR
+            x = self.conv_first(x)
+
+            res = self.forward_features(x)
+            if not torch.is_tensor(res):
+                res, loss_moe = res
+
+            x = self.conv_after_body(res) + x
+            x = self.upsample(x)
+        elif self.upsampler == 'nearest+conv':
+            # for real-world SR
+            x = self.conv_first(x)
+
+            res = self.forward_features(x)
+            if not torch.is_tensor(res):
+                res, loss_moe = res
+
+            x = self.conv_after_body(res) + x
+            x = self.conv_before_upsample(x)
+            x = self.lrelu(self.conv_up1(torch.nn.functional.interpolate(x, scale_factor=2, mode='nearest')))
+            x = self.lrelu(self.conv_up2(torch.nn.functional.interpolate(x, scale_factor=2, mode='nearest')))
+            x = self.conv_last(self.lrelu(self.conv_hr(x)))
+        else:
+            # for image denoising and JPEG compression artifact reduction
+            x_first = self.conv_first(x)
+
+            res = self.forward_features(x_first)
+            if not torch.is_tensor(res):
+                res, loss_moe = res
+
+            res = self.conv_after_body(res) + x_first
+            x = x + self.conv_last(res)
+
+        x = x / self.img_range + self.mean
+        if self.upsampler == "pixelshuffle_aux":
+            return x[:, :, :H*self.upscale, :W*self.upscale], aux, loss_moe
+
+        elif self.upsampler == "pixelshuffle_hf":
+            x_out = x_out / self.img_range + self.mean
+            return x_out[:, :, :H*self.upscale, :W*self.upscale], x[:, :, :H*self.upscale, :W*self.upscale], x_hf[:, :, :H*self.upscale, :W*self.upscale], loss_moe
+
+        else:
+            return x[:, :, :H*self.upscale, :W*self.upscale], loss_moe
+
+    def flops(self):
+        flops = 0
+        H, W = self.patches_resolution
+        flops += H * W * 3 * self.embed_dim * 9
+        flops += self.patch_embed.flops()
+        for i, layer in enumerate(self.layers):
+            flops += layer.flops()
+        flops += H * W * 3 * self.embed_dim * self.embed_dim
+        flops += self.upsample.flops()
+        return flops

swin2_mose/moe.py

@@ -0,0 +1,323 @@
+#
+# Source code: https://github.com/davidmrau/mixture-of-experts
+#
+
+# Sparsely-Gated Mixture-of-Experts Layers.
+# See "Outrageously Large Neural Networks"
+# https://arxiv.org/abs/1701.06538
+#
+# Author: David Rau
+#
+# The code is based on the TensorFlow implementation:
+# https://github.com/tensorflow/tensor2tensor/blob/master/tensor2tensor/utils/expert_utils.py
+
+
+import torch
+import torch.nn as nn
+from torch.distributions.normal import Normal
+from copy import deepcopy
+import numpy as np
+
+from utils import Mlp as MLP
+
+class SparseDispatcher(object):
+    """Helper for implementing a mixture of experts.
+    The purpose of this class is to create input minibatches for the
+    experts and to combine the results of the experts to form a unified
+    output tensor.
+    There are two functions:
+    dispatch - take an input Tensor and create input Tensors for each expert.
+    combine - take output Tensors from each expert and form a combined output
+      Tensor.  Outputs from different experts for the same batch element are
+      summed together, weighted by the provided "gates".
+    The class is initialized with a "gates" Tensor, which specifies which
+    batch elements go to which experts, and the weights to use when combining
+    the outputs.  Batch element b is sent to expert e iff gates[b, e] != 0.
+    The inputs and outputs are all two-dimensional [batch, depth].
+    Caller is responsible for collapsing additional dimensions prior to
+    calling this class and reshaping the output to the original shape.
+    See common_layers.reshape_like().
+    Example use:
+    gates: a float32 `Tensor` with shape `[batch_size, num_experts]`
+    inputs: a float32 `Tensor` with shape `[batch_size, input_size]`
+    experts: a list of length `num_experts` containing sub-networks.
+    dispatcher = SparseDispatcher(num_experts, gates)
+    expert_inputs = dispatcher.dispatch(inputs)
+    expert_outputs = [experts[i](expert_inputs[i]) for i in range(num_experts)]
+    outputs = dispatcher.combine(expert_outputs)
+    The preceding code sets the output for a particular example b to:
+    output[b] = Sum_i(gates[b, i] * experts[i](inputs[b]))
+    This class takes advantage of sparsity in the gate matrix by including in the
+    `Tensor`s for expert i only the batch elements for which `gates[b, i] > 0`.
+    """
+
+    def __init__(self, num_experts, gates):
+        """Create a SparseDispatcher."""
+
+        self._gates = gates
+        self._num_experts = num_experts
+        # sort experts
+        sorted_experts, index_sorted_experts = torch.nonzero(gates).sort(0)
+        # drop indices
+        _, self._expert_index = sorted_experts.split(1, dim=1)
+        # get according batch index for each expert
+        self._batch_index = torch.nonzero(gates)[index_sorted_experts[:, 1], 0]
+        # calculate num samples that each expert gets
+        self._part_sizes = (gates > 0).sum(0).tolist()
+        # expand gates to match with self._batch_index
+        gates_exp = gates[self._batch_index.flatten()]
+        self._nonzero_gates = torch.gather(gates_exp, 1, self._expert_index)
+
+    def dispatch(self, inp):
+        """Create one input Tensor for each expert.
+        The `Tensor` for a expert `i` contains the slices of `inp` corresponding
+        to the batch elements `b` where `gates[b, i] > 0`.
+        Args:
+          inp: a `Tensor` of shape "[batch_size, <extra_input_dims>]`
+        Returns:
+          a list of `num_experts` `Tensor`s with shapes
+            `[expert_batch_size_i, <extra_input_dims>]`.
+        """
+
+        # assigns samples to experts whose gate is nonzero
+
+        # expand according to batch index so we can just split by _part_sizes
+        inp_exp = inp[self._batch_index].squeeze(1)
+        return torch.split(inp_exp, self._part_sizes, dim=0)
+
+    def combine(self, expert_out, multiply_by_gates=True, cnn_combine=None):
+        """Sum together the expert output, weighted by the gates.
+        The slice corresponding to a particular batch element `b` is computed
+        as the sum over all experts `i` of the expert output, weighted by the
+        corresponding gate values.  If `multiply_by_gates` is set to False, the
+        gate values are ignored.
+        Args:
+          expert_out: a list of `num_experts` `Tensor`s, each with shape
+            `[expert_batch_size_i, <extra_output_dims>]`.
+          multiply_by_gates: a boolean
+        Returns:
+          a `Tensor` with shape `[batch_size, <extra_output_dims>]`.
+        """
+        # apply exp to expert outputs, so we are not longer in log space
+        stitched = torch.cat(expert_out, 0)
+
+        if multiply_by_gates:
+            stitched = stitched.mul(self._nonzero_gates.unsqueeze(1))
+        zeros = torch.zeros((self._gates.size(0),) + expert_out[-1].shape[1:],
+                            requires_grad=True, device=stitched.device)
+        # combine samples that have been processed by the same k experts
+
+        if cnn_combine is not None:
+            return self.smartly_combine(stitched, cnn_combine)
+
+        combined = zeros.index_add(0, self._batch_index, stitched.float())
+        return combined
+
+    def smartly_combine(self, stitched, cnn_combine):
+        idxes = []
+        for i in self._batch_index.unique():
+            idx = (self._batch_index == i).nonzero().squeeze(1)
+            idxes.append(idx)
+        idxes = torch.stack(idxes)
+        return cnn_combine(stitched[idxes]).squeeze(1)
+
+    def expert_to_gates(self):
+        """Gate values corresponding to the examples in the per-expert `Tensor`s.
+        Returns:
+          a list of `num_experts` one-dimensional `Tensor`s with type `tf.float32`
+              and shapes `[expert_batch_size_i]`
+        """
+        # split nonzero gates for each expert
+        return torch.split(self._nonzero_gates, self._part_sizes, dim=0)
+
+
+def build_experts(experts_cfg, default_cfg, num_experts):
+    experts_cfg = deepcopy(experts_cfg)
+    if experts_cfg is None:
+        # old build way
+        return nn.ModuleList([
+            MLP(*default_cfg)
+            for i in range(num_experts)])
+    # new build way: mix mlp with leff
+    experts = []
+    for e_cfg in experts_cfg:
+        type_ = e_cfg.pop('type')
+        if type_ == 'mlp':
+            experts.append(MLP(*default_cfg))
+    return nn.ModuleList(experts)
+
+
+class MoE(nn.Module):
+    """Call a Sparsely gated mixture of experts layer with 1-layer
+       Feed-Forward networks as experts.
+
+    Args:
+    input_size: integer - size of the input
+    output_size: integer - size of the input
+    num_experts: an integer - number of experts
+    hidden_size: an integer - hidden size of the experts
+    noisy_gating: a boolean
+    k: an integer - how many experts to use for each batch element
+    """
+
+    def __init__(self, input_size, output_size, num_experts, hidden_size,
+                 experts=None, noisy_gating=True, k=4,
+                 x_gating=None, with_noise=True, with_smart_merger=None):
+        super(MoE, self).__init__()
+        self.noisy_gating = noisy_gating
+        self.num_experts = num_experts
+        self.output_size = output_size
+        self.input_size = input_size
+        self.hidden_size = hidden_size
+        self.k = k
+        self.with_noise = with_noise
+        # instantiate experts
+        self.experts = build_experts(
+            experts,
+            (self.input_size, self.hidden_size, self.output_size),
+            num_experts)
+        self.w_gate = nn.Parameter(torch.zeros(input_size, num_experts), requires_grad=True)
+        self.w_noise = nn.Parameter(torch.zeros(input_size, num_experts), requires_grad=True)
+
+        self.x_gating = x_gating
+        if self.x_gating == 'conv1d':
+            self.x_gate = nn.Conv1d(4096, 1, kernel_size=3, padding=1)
+
+        self.softplus = nn.Softplus()
+        self.softmax = nn.Softmax(1)
+        self.register_buffer("mean", torch.tensor([0.0]))
+        self.register_buffer("std", torch.tensor([1.0]))
+        assert(self.k <= self.num_experts)
+
+        self.cnn_combine = None
+        if with_smart_merger == 'v1':
+            print('with SMART MERGER')
+            self.cnn_combine = nn.Conv2d(self.k, 1, kernel_size=3, padding=1)
+
+    def cv_squared(self, x):
+        """The squared coefficient of variation of a sample.
+        Useful as a loss to encourage a positive distribution to be more uniform.
+        Epsilons added for numerical stability.
+        Returns 0 for an empty Tensor.
+        Args:
+        x: a `Tensor`.
+        Returns:
+        a `Scalar`.
+        """
+        eps = 1e-10
+        # if only num_experts = 1
+
+        if x.shape[0] == 1:
+            return torch.tensor([0], device=x.device, dtype=x.dtype)
+        return x.float().var() / (x.float().mean()**2 + eps)
+
+    def _gates_to_load(self, gates):
+        """Compute the true load per expert, given the gates.
+        The load is the number of examples for which the corresponding gate is >0.
+        Args:
+        gates: a `Tensor` of shape [batch_size, n]
+        Returns:
+        a float32 `Tensor` of shape [n]
+        """
+        return (gates > 0).sum(0)
+
+    def _prob_in_top_k(self, clean_values, noisy_values, noise_stddev, noisy_top_values):
+        """Helper function to NoisyTopKGating.
+        Computes the probability that value is in top k, given different random noise.
+        This gives us a way of backpropagating from a loss that balances the number
+        of times each expert is in the top k experts per example.
+        In the case of no noise, pass in None for noise_stddev, and the result will
+        not be differentiable.
+        Args:
+        clean_values: a `Tensor` of shape [batch, n].
+        noisy_values: a `Tensor` of shape [batch, n].  Equal to clean values plus
+          normally distributed noise with standard deviation noise_stddev.
+        noise_stddev: a `Tensor` of shape [batch, n], or None
+        noisy_top_values: a `Tensor` of shape [batch, m].
+           "values" Output of tf.top_k(noisy_top_values, m).  m >= k+1
+        Returns:
+        a `Tensor` of shape [batch, n].
+        """
+        batch = clean_values.size(0)
+        m = noisy_top_values.size(1)
+        top_values_flat = noisy_top_values.flatten()
+
+        threshold_positions_if_in = torch.arange(batch, device=clean_values.device) * m + self.k
+        threshold_if_in = torch.unsqueeze(torch.gather(top_values_flat, 0, threshold_positions_if_in), 1)
+        is_in = torch.gt(noisy_values, threshold_if_in)
+        threshold_positions_if_out = threshold_positions_if_in - 1
+        threshold_if_out = torch.unsqueeze(torch.gather(top_values_flat, 0, threshold_positions_if_out), 1)
+        # is each value currently in the top k.
+        normal = Normal(self.mean, self.std)
+        prob_if_in = normal.cdf((clean_values - threshold_if_in)/noise_stddev)
+        prob_if_out = normal.cdf((clean_values - threshold_if_out)/noise_stddev)
+        prob = torch.where(is_in, prob_if_in, prob_if_out)
+        return prob
+
+    def noisy_top_k_gating(self, x, train, noise_epsilon=1e-2):
+        """Noisy top-k gating.
+          See paper: https://arxiv.org/abs/1701.06538.
+          Args:
+            x: input Tensor with shape [batch_size, input_size]
+            train: a boolean - we only add noise at training time.
+            noise_epsilon: a float
+          Returns:
+            gates: a Tensor with shape [batch_size, num_experts]
+            load: a Tensor with shape [num_experts]
+        """
+        clean_logits = x @ self.w_gate
+        if self.noisy_gating and train:
+            raw_noise_stddev = x @ self.w_noise
+            noise_stddev = ((self.softplus(raw_noise_stddev) + noise_epsilon))
+            noisy_logits = clean_logits + (torch.randn_like(clean_logits) * noise_stddev)
+            logits = noisy_logits
+        else:
+            logits = clean_logits
+
+        # calculate topk + 1 that will be needed for the noisy gates
+        top_logits, top_indices = logits.topk(min(self.k + 1, self.num_experts), dim=1)
+        top_k_logits = top_logits[:, :self.k]
+        top_k_indices = top_indices[:, :self.k]
+        top_k_gates = self.softmax(top_k_logits)
+
+        zeros = torch.zeros_like(logits, requires_grad=True)
+        gates = zeros.scatter(1, top_k_indices, top_k_gates)
+
+        if self.noisy_gating and self.k < self.num_experts and train:
+            load = (self._prob_in_top_k(clean_logits, noisy_logits, noise_stddev, top_logits)).sum(0)
+        else:
+            load = self._gates_to_load(gates)
+        return gates, load
+
+    def forward(self, x, loss_coef=1e-2):
+        """Args:
+        x: tensor shape [batch_size, input_size]
+        train: a boolean scalar.
+        loss_coef: a scalar - multiplier on load-balancing losses
+
+        Returns:
+        y: a tensor with shape [batch_size, output_size].
+        extra_training_loss: a scalar.  This should be added into the overall
+        training loss of the model.  The backpropagation of this loss
+        encourages all experts to be approximately equally used across a batch.
+        """
+        if self.x_gating is not None:
+            xg = self.x_gate(x).squeeze(1)
+        else:
+            xg = x.mean(1)
+
+        gates, load = self.noisy_top_k_gating(
+            xg, self.training and self.with_noise)
+        # calculate importance loss
+        importance = gates.sum(0)
+        #
+        loss = self.cv_squared(importance) + self.cv_squared(load)
+        loss *= loss_coef
+
+        dispatcher = SparseDispatcher(self.num_experts, gates)
+        expert_inputs = dispatcher.dispatch(x)
+        gates = dispatcher.expert_to_gates()
+        expert_outputs = [self.experts[i](expert_inputs[i])
+                          for i in range(self.num_experts)]
+        y = dispatcher.combine(expert_outputs, cnn_combine=self.cnn_combine)
+        return y, loss

swin2_mose/run.py

@@ -0,0 +1,20 @@
+import torch
+from model import Swin2SR
+
+model_weights = "model-70.pt"
+model_params = {
+    "upscale": 2,
+    "in_chans": 4,
+    "img_size": 64,
+    "window_size": 16,
+    "img_range": 1.,
+    "depths": [6, 6, 6, 6],
+    "embed_dim": 90,
+    "num_heads": [6, 6, 6, 6],
+    "mlp_ratio": 2,
+    "upsampler": "pixelshuffledirect",
+    "resi_connection": "1conv"
+}
+
+sr_model = Swin2SR(**model_params)
+sr_model.load_state_dict(torch.load(model_weights))

swin2_mose/utils.py

@@ -0,0 +1,56 @@
+from torch import nn
+
+
+def window_reverse(windows, window_size, H, W):
+    """
+    Args:
+        windows: (num_windows*B, window_size, window_size, C)
+        window_size (int): Window size
+        H (int): Height of image
+        W (int): Width of image
+
+    Returns:
+        x: (B, H, W, C)
+    """
+    B = int(windows.shape[0] / (H * W / window_size / window_size))
+    x = windows.view(B, H // window_size, W // window_size, window_size,
+                     window_size, -1)
+    x = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(B, H, W, -1)
+    return x
+
+
+class Mlp(nn.Module):
+    def __init__(self, in_features, hidden_features=None, out_features=None,
+                 act_layer=nn.GELU, drop=0.):
+        super().__init__()
+        out_features = out_features or in_features
+        hidden_features = hidden_features or in_features
+        self.fc1 = nn.Linear(in_features, hidden_features)
+        self.act = act_layer()
+        self.fc2 = nn.Linear(hidden_features, out_features)
+        self.drop = nn.Dropout(drop)
+
+    def forward(self, x):
+        x = self.fc1(x)
+        x = self.act(x)
+        x = self.drop(x)
+        x = self.fc2(x)
+        x = self.drop(x)
+        return x
+
+
+def window_partition(x, window_size):
+    """
+    Args:
+        x: (B, H, W, C)
+        window_size (int): window size
+
+    Returns:
+        windows: (num_windows*B, window_size, window_size, C)
+    """
+    B, H, W, C = x.shape
+    x = x.view(B, H // window_size, window_size,
+               W // window_size, window_size, C)
+    windows = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(
+        -1, window_size, window_size, C)
+    return windows

swin2_mose/weights/model-70.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c9f1229521879af2c8162f7a32fe278e487d0bc0826dddccc87a4e22294aa067
+size 118890958