try gtp vit

main.py

@@ -478,11 +478,7 @@ class Main(object):
                 batch, 'train')
 
             pred = self.model.forward(sub, rel, neg_ent, self.p.train_strategy)
-            try:
+            loss = self.model.loss(pred, label, sub_samp)
-                loss = self.model.loss(pred, label, sub_samp)
-            except Exception as e:
-                print(pred)
-                raise e
 
             loss.backward()
             self.optimizer.step()

models.py

@@ -1,16 +1,17 @@
 import torch
-from torch import nn, einsum
+from torch import nn
 import torch.nn.functional as F
 import numpy as np
 from functools import partial
 from einops.layers.torch import Rearrange, Reduce
-from einops import rearrange, repeat
 from utils import *
 from layers import *
 from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD
 from timm.models.layers import DropPath, trunc_normal_
 from timm.models.registry import register_model
 from timm.layers.helpers import to_2tuple
+from typing import *
+import math
 
 
 class ConvE(torch.nn.Module):
@@ -527,6 +528,22 @@ class FouriER(torch.nn.Module):
 
         self.network = nn.ModuleList(network)
         self.norm = norm_layer(embed_dims[-1])
+        self.graph_type = 'Spatial'
+        N = (image_h // patch_size)**2
+        if self.graph_type in ["Spatial", "Mixed"]:
+            # Create a range tensor of node indices
+            indices = torch.arange(N)
+            # Reshape the indices tensor to create a grid of row and column indices
+            row_indices = indices.view(-1, 1).expand(-1, N)
+            col_indices = indices.view(1, -1).expand(N, -1)
+            # Compute the adjacency matrix
+            row1, col1 = row_indices // int(math.sqrt(N)), row_indices % int(math.sqrt(N))
+            row2, col2 = col_indices // int(math.sqrt(N)), col_indices % int(math.sqrt(N))
+            graph = ((abs(row1 - row2) <= 1).float() * (abs(col1 - col2) <= 1).float())
+            graph = graph - torch.eye(N)
+            self.spatial_graph = graph.cuda() # comment .to("cuda") if the environment is cpu
+        self.class_token = False
+        self.token_scale = False
         self.head = nn.Linear(
                 embed_dims[-1], num_classes) if num_classes > 0 \
                 else nn.Identity()
@@ -544,8 +561,45 @@ class FouriER(torch.nn.Module):
 
     def forward_tokens(self, x):
         outs = []
+        B, C, H, W = x.shape
+        N = H*W
+        if self.graph_type in ["Semantic", "Mixed"]:
+            # Generate the semantic graph w.r.t. the cosine similarity between tokens
+            # Compute cosine similarity
+            if self.class_token:
+                x_normed = x[:, 1:] / x[:, 1:].norm(dim=-1, keepdim=True)
+            else:
+                x_normed = x / x.norm(dim=-1, keepdim=True)
+            x_cossim = x_normed @ x_normed.transpose(-1, -2)
+            threshold = torch.kthvalue(x_cossim, N-1-self.num_neighbours, dim=-1, keepdim=True)[0] # B,H,1,1 
+            semantic_graph = torch.where(x_cossim>=threshold, 1.0, 0.0)
+            if self.class_token:
+                semantic_graph = semantic_graph - torch.eye(N-1, device=semantic_graph.device).unsqueeze(0)
+            else:
+                semantic_graph = semantic_graph - torch.eye(N, device=semantic_graph.device).unsqueeze(0)
+        
+        if self.graph_type == "None":
+            graph = None
+        else:
+            if self.graph_type == "Spatial":
+                graph = self.spatial_graph.unsqueeze(0).expand(B,-1,-1)#.to(x.device)
+            elif self.graph_type == "Semantic":
+                graph = semantic_graph
+            elif self.graph_type == "Mixed":
+                # Integrate the spatial graph and semantic graph
+                spatial_graph = self.spatial_graph.unsqueeze(0).expand(B,-1,-1).to(x.device)
+                graph = torch.bitwise_or(semantic_graph.int(), spatial_graph.int()).float()
+            
+            # Symmetrically normalize the graph
+            degree = graph.sum(-1) # B, N
+            degree = torch.diag_embed(degree**(-1/2))
+            graph = degree @ graph @ degree
+
         for idx, block in enumerate(self.network):
-            x = block(x)
+            try:
+                x = block(x, graph)
+            except:
+                x = block(x)
         # output only the features of last layer for image classification
         return x
 
@@ -558,8 +612,6 @@ class FouriER(torch.nn.Module):
         z = self.forward_embeddings(y)
         z = self.forward_tokens(z)
         z = z.mean([-2, -1])
-        if np.count_nonzero(np.isnan(z)) > 0:
-            print("ZZZ")
         z = self.norm(z)
         x = self.head(z)
         x = self.hidden_drop(x)
@@ -706,7 +758,7 @@ def basic_blocks(dim, index, layers,
             use_layer_scale=use_layer_scale, 
             layer_scale_init_value=layer_scale_init_value, 
             ))
-    blocks = nn.Sequential(*blocks)
+    blocks = SeqModel(*blocks)
 
     return blocks
 
@@ -871,202 +923,278 @@ def window_reverse(windows, window_size, H, W):
     x = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(B, -1, H, W)
     return x
 
-def cast_tuple(val, length = 1):
+class SeqModel(nn.Sequential):
-    return val if isinstance(val, tuple) else ((val,) * length)
+	def forward(self, *inputs):
+		for module in self._modules.values():
+			if type(inputs) == tuple:
+				inputs = module(*inputs)
+			else:
+				inputs = module(inputs)
+		return inputs
 
-# helper classes
+def propagate(x: torch.Tensor, weight: torch.Tensor, 
-
+              index_kept: torch.Tensor, index_prop: torch.Tensor, 
-class ChanLayerNorm(nn.Module):
+              standard: str = "None", alpha: Optional[float] = 0, 
-    def __init__(self, dim, eps = 1e-5):
+              token_scales: Optional[torch.Tensor] = None,
-        super().__init__()
+              cls_token=True):
-        self.eps = eps
+    """
-        self.g = nn.Parameter(torch.ones(1, dim, 1, 1))
+    Propagate tokens based on the selection results.
-        self.b = nn.Parameter(torch.zeros(1, dim, 1, 1))
+    ================================================
-
+    Args:
-    def forward(self, x):
+        - x: Tensor([B, N, C]): the feature map of N tokens, including the [CLS] token.
-        var = torch.var(x, dim = 1, unbiased = False, keepdim = True)
+        
-        mean = torch.mean(x, dim = 1, keepdim = True)
+        - weight: Tensor([B, N-1, N-1]): the weight of each token propagated to the other tokens, 
-        return (x - mean) / (var + self.eps).sqrt() * self.g + self.b
+                                         excluding the [CLS] token. weight could be a pre-defined 
-
+                                         graph of the current feature map (by default) or the
-class OverlappingPatchEmbed(nn.Module):
+                                         attention map (need to manually modify the Block Module).
-    def __init__(self, dim_in, dim_out, stride = 2):
+                                         
-        super().__init__()
+        - index_kept: Tensor([B, N-1-num_prop]): the index of kept image tokens in the feature map X
-        kernel_size = stride * 2 - 1
+        
-        padding = kernel_size // 2
+        - index_prop: Tensor([B, num_prop]): the index of propagated image tokens in the feature map X
-        self.conv = nn.Conv2d(dim_in, dim_out, kernel_size, stride = stride, padding = padding)
+        
-
+        - standard: str: the method applied to propagate the tokens, including "None", "Mean" and 
-    def forward(self, x):
+                         "GraphProp"
-        return self.conv(x)
+        
-
+        - alpha: float: the coefficient of propagated features
-class PEG(nn.Module):
+        
-    def __init__(self, dim, kernel_size = 3):
+        - token_scales: Tensor([B, N]): the scale of tokens, including the [CLS] token. token_scales
-        super().__init__()
+                                        is None by default. If it is not None, then token_scales 
-        self.proj = nn.Conv2d(dim, dim, kernel_size = kernel_size, padding = kernel_size // 2, groups = dim, stride = 1)
+                                        represents the scales of each token and should sum up to N.
-
+        
-    def forward(self, x):
+    Return:
-        return self.proj(x) + x
+        - x: Tensor([B, N-1-num_prop, C]): the feature map after propagation
-
+        
-# feedforward
+        - weight: Tensor([B, N-1-num_prop, N-1-num_prop]): the graph of feature map after propagation
-
+        
-class FeedForwardDSSA(nn.Module):
+        - token_scales: Tensor([B, N-1-num_prop]): the scale of tokens after propagation
-    def __init__(self, dim, mult = 4, dropout = 0.):
+    """
-        super().__init__()
+    
-        inner_dim = int(dim * mult)
+    B, N, C = x.shape
-        self.net = nn.Sequential(
+    
-            ChanLayerNorm(dim),
+    # Step 1: divide tokens
-            nn.Conv2d(dim, inner_dim, 1),
+    if cls_token:
-            nn.GELU(),
+        x_cls = x[:, 0:1] # B, 1, C
-            nn.Dropout(dropout),
+    x_kept = x.gather(dim=1, index=index_kept.unsqueeze(-1).expand(-1,-1,C)) # B, N-1-num_prop, C
-            nn.Conv2d(inner_dim, dim, 1),
+    x_prop = x.gather(dim=1, index=index_prop.unsqueeze(-1).expand(-1,-1,C)) # B, num_prop, C
-            nn.Dropout(dropout)
+    
-        )
+    # Step 2: divide token_scales if it is not None
-    def forward(self, x):
+    if token_scales is not None:
-        return self.net(x)
+        if cls_token:
-
+            token_scales_cls = token_scales[:, 0:1] # B, 1
-# attention
+        token_scales_kept = token_scales.gather(dim=1, index=index_kept) # B, N-1-num_prop
-
+        token_scales_prop = token_scales.gather(dim=1, index=index_prop) # B, num_prop
-class DSSA(nn.Module):
+    
-    def __init__(
+    # Step 3: propagate tokens
-        self,
+    if standard == "None":
-        dim,
-        heads = 8,
-        dim_head = 32,
-        dropout = 0.,
-        window_size = 7
-    ):
-        super().__init__()
-        self.heads = heads
-        self.scale = dim_head ** -0.5
-        self.window_size = window_size
-        inner_dim = dim_head * heads
-
-        self.norm = ChanLayerNorm(dim)
-
-        self.attend = nn.Sequential(
-            nn.Softmax(dim = -1),
-            nn.Dropout(dropout)
-        )
-
-        self.to_qkv = nn.Conv1d(dim, inner_dim * 3, 1, bias = False)
-
-        # window tokens
-
-        self.window_tokens = nn.Parameter(torch.randn(dim))
-
-        # prenorm and non-linearity for window tokens
-        # then projection to queries and keys for window tokens
-
-        self.window_tokens_to_qk = nn.Sequential(
-            nn.LayerNorm(dim_head),
-            nn.GELU(),
-            Rearrange('b h n c -> b (h c) n'),
-            nn.Conv1d(inner_dim, inner_dim * 2, 1),
-            Rearrange('b (h c) n -> b h n c', h = heads),
-        )
-
-        # window attention
-
-        self.window_attend = nn.Sequential(
-            nn.Softmax(dim = -1),
-            nn.Dropout(dropout)
-        )
-
-        self.to_out = nn.Sequential(
-            nn.Conv2d(inner_dim, dim, 1),
-            nn.Dropout(dropout)
-        )
-
-    def forward(self, x):
         """
-        einstein notation
+        No further propagation
-
-        b - batch
-        c - channels
-        w1 - window size (height)
-        w2 - also window size (width)
-        i - sequence dimension (source)
-        j - sequence dimension (target dimension to be reduced)
-        h - heads
-        x - height of feature map divided by window size
-        y - width of feature map divided by window size
         """
+        pass
+        
+    elif standard == "Mean":
+        """
+        Calculate the mean of all the propagated tokens,
+        and concatenate the result token back to kept tokens.
+        """
+        # naive average
+        x_prop = x_prop.mean(1, keepdim=True) # B, 1, C
+        # Concatenate the average token 
+        x_kept = torch.cat((x_kept, x_prop), dim=1) # B, N-num_prop, C
+            
+    elif standard == "GraphProp":
+        """
+        Propagate all the propagated token to kept token
+        with respect to the weights and token scales.
+        """
+        assert weight is not None, "The graph weight is needed for graph propagation"
+        
+        # Step 3.1: divide propagation weights.
+        if cls_token:
+            index_kept = index_kept - 1 # since weights do not include the [CLS] token
+            index_prop = index_prop - 1 # since weights do not include the [CLS] token
+            weight = weight.gather(dim=1, index=index_kept.unsqueeze(-1).expand(-1,-1,N-1)) # B, N-1-num_prop, N-1
+            weight_prop = weight.gather(dim=2, index=index_prop.unsqueeze(1).expand(-1,weight.shape[1],-1)) # B, N-1-num_prop, num_prop
+            weight = weight.gather(dim=2, index=index_kept.unsqueeze(1).expand(-1,weight.shape[1],-1)) # B, N-1-num_prop, N-1-num_prop
+        else:
+            weight = weight.gather(dim=1, index=index_kept.unsqueeze(-1).expand(-1,-1,N)) # B, N-1-num_prop, N-1
+            weight_prop = weight.gather(dim=2, index=index_prop.unsqueeze(1).expand(-1,weight.shape[1],-1)) # B, N-1-num_prop, num_prop
+            weight = weight.gather(dim=2, index=index_kept.unsqueeze(1).expand(-1,weight.shape[1],-1)) # B, N-1-num_prop, N-1-num_prop
+        
+        # Step 3.2: generate the broadcast message and propagate the message to corresponding kept tokens
+        # Simple implementation
+        x_prop = weight_prop @ x_prop # B, N-1-num_prop, C
+        x_kept = x_kept + alpha * x_prop # B, N-1-num_prop, C
+        
+        """ scatter_reduce implementation for batched inputs
+        # Get the non-zero values
+        non_zero_indices = torch.nonzero(weight_prop, as_tuple=True)
+        non_zero_values = weight_prop[non_zero_indices]
+        
+        # Sparse multiplication
+        batch_indices, row_indices, col_indices = non_zero_indices
+        sparse_matmul = alpha * non_zero_values[:, None] * x_prop[batch_indices, col_indices, :]
+        reduce_indices = batch_indices * x_kept.shape[1] + row_indices
+        
+        x_kept = x_kept.reshape(-1, C).scatter_reduce(dim=0, 
+                                                      index=reduce_indices[:, None], 
+                                                      src=sparse_matmul, 
+                                                      reduce="sum",
+                                                      include_self=True)
+        x_kept = x_kept.reshape(B, -1, C)
+        """
+        
+        # Step 3.3: calculate the scale of each token if token_scales is not None
+        if token_scales is not None:
+            if cls_token:
+                token_scales_cls = token_scales[:, 0:1] # B, 1
+                token_scales = token_scales[:, 1:]
+            token_scales_kept = token_scales.gather(dim=1, index=index_kept) # B, N-1-num_prop
+            token_scales_prop = token_scales.gather(dim=1, index=index_prop) # B, num_prop
+            token_scales_prop = weight_prop @ token_scales_prop.unsqueeze(-1) # B, N-1-num_prop, 1
+            token_scales = token_scales_kept + alpha * token_scales_prop.squeeze(-1) # B, N-1-num_prop
+            if cls_token:
+                token_scales = torch.cat((token_scales_cls, token_scales), dim=1) # B, N-num_prop
+    else:
+        assert False, "Propagation method \'%f\' has not been supported yet." % standard
+    
+    
+    if cls_token:
+        # Step 4： concatenate the [CLS] token and generate returned value
+        x = torch.cat((x_cls, x_kept), dim=1) # B, N-num_prop, C
+    else:
+        x = x_kept
+    return x, weight, token_scales
 
-        batch, height, width, heads, wsz = x.shape[0], *x.shape[-2:], self.heads, self.window_size
-        assert (height % wsz) == 0 and (width % wsz) == 0, f'height {height} and width {width} must be divisible by window size {wsz}'
-        num_windows = (height // wsz) * (width // wsz)
 
-        x = self.norm(x)
 
-        # fold in windows for "depthwise" attention - not sure why it is named depthwise when it is just "windowed" attention
+def select(weight: torch.Tensor, standard: str = "None", num_prop: int = 0, cls_token = True):
-
+    """
-        x = rearrange(x, 'b c (h w1) (w w2) -> (b h w) c (w1 w2)', w1 = wsz, w2 = wsz)
+    Select image tokens to be propagated. The [CLS] token will be ignored. 
-
+    ======================================================================
-        # add windowing tokens
+    Args:
-
+        - weight: Tensor([B, H, N, N]): used for selecting the kept tokens. Only support the
-        w = repeat(self.window_tokens, 'c -> b c 1', b = x.shape[0])
+                                        attention map of tokens at the moment.
-        x = torch.cat((w, x), dim = -1)
+        
-
+        - standard: str: the method applied to select the tokens
-        # project for queries, keys, value
+        
-
+        - num_prop: int: the number of tokens to be propagated
-        q, k, v = self.to_qkv(x).chunk(3, dim = 1)
+        
-
+    Return:
-        # split out heads
+        - index_kept: Tensor([B, N-1-num_prop]): the index of kept tokens 
-
+        
-        q, k, v = map(lambda t: rearrange(t, 'b (h d) ... -> b h (...) d', h = heads), (q, k, v))
+        - index_prop: Tensor([B, num_prop]): the index of propagated tokens
-
+    """
-        # scale
+    
-
+    assert len(weight.shape) == 4, "Selection methods on tensors other than the attention map haven't been supported yet."
-        q = q * self.scale
+    B, H, N1, N2 = weight.shape
-
+    assert N1 == N2, "Selection methods on tensors other than the attention map haven't been supported yet."
-        # similarity
+    N = N1
-
+    assert num_prop >= 0, "The number of propagated/pruned tokens must be non-negative."
-        dots = einsum('b h i d, b h j d -> b h i j', q, k)
+    
-
+    if cls_token:
-        # attention
+        if standard == "CLSAttnMean":
-
+            token_rank = weight[:,:,0,1:].mean(1)
-        attn = self.attend(dots)
+            
-
+        elif standard == "CLSAttnMax":
-        # aggregate values
+            token_rank = weight[:,:,0,1:].max(1)[0]
-
+                
-        out = torch.matmul(attn, v)
+        elif standard == "IMGAttnMean":
-
+            token_rank = weight[:,:,:,1:].sum(-2).mean(1)
-        # split out windowed tokens
+        
-
+        elif standard == "IMGAttnMax":
-        window_tokens, windowed_fmaps = out[:, :, 0], out[:, :, 1:]
+            token_rank = weight[:,:,:,1:].sum(-2).max(1)[0]
-
+                
-        # early return if there is only 1 window
+        elif standard == "DiagAttnMean":
-
+            token_rank = torch.diagonal(weight, dim1=-2, dim2=-1)[:,:,1:].mean(1)
-        if num_windows == 1:
+            
-            fmap = rearrange(windowed_fmaps, '(b x y) h (w1 w2) d -> b (h d) (x w1) (y w2)', x = height // wsz, y = width // wsz, w1 = wsz, w2 = wsz)
+        elif standard == "DiagAttnMax":
-            return self.to_out(fmap)
+            token_rank = torch.diagonal(weight, dim1=-2, dim2=-1)[:,:,1:].max(1)[0]
-
+            
-        # carry out the pointwise attention, the main novelty in the paper
+        elif standard == "MixedAttnMean":
-
+            token_rank_1 = torch.diagonal(weight, dim1=-2, dim2=-1)[:,:,1:].mean(1)
-        window_tokens = rearrange(window_tokens, '(b x y) h d -> b h (x y) d', x = height // wsz, y = width // wsz)
+            token_rank_2 = weight[:,:,:,1:].sum(-2).mean(1)
-        windowed_fmaps = rearrange(windowed_fmaps, '(b x y) h n d -> b h (x y) n d', x = height // wsz, y = width // wsz)
+            token_rank = token_rank_1 * token_rank_2
-
+            
-        # windowed queries and keys (preceded by prenorm activation)
+        elif standard == "MixedAttnMax":
-
+            token_rank_1 = torch.diagonal(weight, dim1=-2, dim2=-1)[:,:,1:].max(1)[0]
-        w_q, w_k = self.window_tokens_to_qk(window_tokens).chunk(2, dim = -1)
+            token_rank_2 = weight[:,:,:,1:].sum(-2).max(1)[0]
-
+            token_rank = token_rank_1 * token_rank_2
-        # scale
+            
-
+        elif standard == "SumAttnMax":
-        w_q = w_q * self.scale
+            token_rank_1 = torch.diagonal(weight, dim1=-2, dim2=-1)[:,:,1:].max(1)[0]
-
+            token_rank_2 = weight[:,:,:,1:].sum(-2).max(1)[0]
-        # similarities
+            token_rank = token_rank_1 + token_rank_2
-
+            
-        w_dots = einsum('b h i d, b h j d -> b h i j', w_q, w_k)
+        elif standard == "CosSimMean":
-
+            weight = weight[:,:,1:,:].mean(1)
-        w_attn = self.window_attend(w_dots)
+            weight = weight / weight.norm(dim=-1, keepdim=True)
-
+            token_rank = -(weight @ weight.transpose(-1, -2)).sum(-1)
-        # aggregate the feature maps from the "depthwise" attention step (the most interesting part of the paper, one i haven't seen before)
+        
-
+        elif standard == "CosSimMax":
-        aggregated_windowed_fmap = einsum('b h i j, b h j w d -> b h i w d', w_attn, windowed_fmaps)
+            weight = weight[:,:,1:,:].max(1)[0]
-
+            weight = weight / weight.norm(dim=-1, keepdim=True)
-        # fold back the windows and then combine heads for aggregation
+            token_rank = -(weight @ weight.transpose(-1, -2)).sum(-1)
-
+            
-        fmap = rearrange(aggregated_windowed_fmap, 'b h (x y) (w1 w2) d -> b (h d) (x w1) (y w2)', x = height // wsz, y = width // wsz, w1 = wsz, w2 = wsz)
+        elif standard == "Random":
-        return self.to_out(fmap)
+            token_rank = torch.randn((B, N-1), device=weight.device)
+                
+        else:
+            print("Type\'", standard, "\' selection not supported.")
+            assert False
+        
+        token_rank = torch.argsort(token_rank, dim=1, descending=True) # B, N-1
+        index_kept = token_rank[:, :-num_prop]+1 # B, N-1-num_prop
+        index_prop = token_rank[:, -num_prop:]+1 # B, num_prop
+            
+    else:
+        if standard == "IMGAttnMean":
+            token_rank = weight.sum(-2).mean(1)
+        
+        elif standard == "IMGAttnMax":
+            token_rank = weight.sum(-2).max(1)[0]
+                
+        elif standard == "DiagAttnMean":
+            token_rank = torch.diagonal(weight, dim1=-2, dim2=-1).mean(1)
+            
+        elif standard == "DiagAttnMax":
+            token_rank = torch.diagonal(weight, dim1=-2, dim2=-1).max(1)[0]
+            
+        elif standard == "MixedAttnMean":
+            token_rank_1 = torch.diagonal(weight, dim1=-2, dim2=-1).mean(1)
+            token_rank_2 = weight.sum(-2).mean(1)
+            token_rank = token_rank_1 * token_rank_2
+            
+        elif standard == "MixedAttnMax":
+            token_rank_1 = torch.diagonal(weight, dim1=-2, dim2=-1).max(1)[0]
+            token_rank_2 = weight.sum(-2).max(1)[0]
+            token_rank = token_rank_1 * token_rank_2
+            
+        elif standard == "SumAttnMax":
+            token_rank_1 = torch.diagonal(weight, dim1=-2, dim2=-1).max(1)[0]
+            token_rank_2 = weight.sum(-2).max(1)[0]
+            token_rank = token_rank_1 + token_rank_2
+            
+        elif standard == "CosSimMean":
+            weight = weight.mean(1)
+            weight = weight / weight.norm(dim=-1, keepdim=True)
+            token_rank = -(weight @ weight.transpose(-1, -2)).sum(-1)
+        
+        elif standard == "CosSimMax":
+            weight = weight.max(1)[0]
+            weight = weight / weight.norm(dim=-1, keepdim=True)
+            token_rank = -(weight @ weight.transpose(-1, -2)).sum(-1)
+            
+        elif standard == "Random":
+            token_rank = torch.randn((B, N-1), device=weight.device)
+                
+        else:
+            print("Type\'", standard, "\' selection not supported.")
+            assert False
+        
+        token_rank = torch.argsort(token_rank, dim=1, descending=True) # B, N-1
+        index_kept = token_rank[:, :-num_prop] # B, N-1-num_prop
+        index_prop = token_rank[:, -num_prop:] # B, num_prop
+    return index_kept, index_prop
 
 class PoolFormerBlock(nn.Module):
     """
@@ -1093,13 +1221,8 @@ class PoolFormerBlock(nn.Module):
         #self.token_mixer = Pooling(pool_size=pool_size)
         # self.token_mixer = FNetBlock()
         self.window_size = 4
-        self.attn_heads = 4
         self.attn_mask = None
-        # self.token_mixer = WindowAttention(dim=dim, window_size=to_2tuple(self.window_size), num_heads=4)
+        self.token_mixer = WindowAttention(dim=dim, window_size=to_2tuple(self.window_size), num_heads=4)
-        self.token_mixer = nn.ModuleList([
-            DSSA(dim, heads=self.attn_heads, window_size=self.window_size),
-            FeedForwardDSSA(dim)
-        ])
         self.norm2 = norm_layer(dim)
         mlp_hidden_dim = int(dim * mlp_ratio)
         self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim, 
@@ -1115,14 +1238,20 @@ class PoolFormerBlock(nn.Module):
             self.layer_scale_2 = nn.Parameter(
                 layer_scale_init_value * torch.ones((dim)), requires_grad=True)
 
-    def forward(self, x):
+    def forward(self, x, weight, token_scales = None):
         B, C, H, W = x.shape
-        # x_windows = window_partition(x, self.window_size)
+        x_windows = window_partition(x, self.window_size)
-        # x_windows = x_windows.view(-1, self.window_size * self.window_size, C)
+        x_windows = x_windows.view(-1, self.window_size * self.window_size, C)
-        # attn_windows = self.token_mixer(x_windows, mask=self.attn_mask)
+        attn_windows = self.token_mixer(x_windows, mask=self.attn_mask)
-        # attn_windows = attn_windows.view(-1, self.window_size, self.window_size, C)
+        attn_windows = attn_windows.view(-1, self.window_size, self.window_size, C)
-        # x_attn = window_reverse(attn_windows, self.window_size, H, W)
+        x_attn = window_reverse(attn_windows, self.window_size, H, W)
-        x_attn = self.token_mixer(x)
+        index_kept, index_prop = select(x_attn, standard="MixedAttnMax", num_prop=0,
+                                            cls_token=False)
+        original_shape = x_attn.shape
+        x_attn = x_attn.view(-1, self.window_size * self.window_size, C)
+        x_attn, weight, token_scales = propagate(x_attn, weight, index_kept, index_prop, standard="GraphProp",
+                                        alpha=0.1, token_scales=token_scales, cls_token=False)
+        x_attn = x_attn.view(*original_shape)
         if self.use_layer_scale:
             x = x + self.drop_path(
                 self.layer_scale_1.unsqueeze(-1).unsqueeze(-1)
@@ -1133,9 +1262,6 @@ class PoolFormerBlock(nn.Module):
         else:
             x = x + self.drop_path(x_attn)
             x = x + self.drop_path(self.mlp(self.norm2(x)))
-
-        if np.count_nonzero(np.isnan(x)) > 0:
-            print("PFBlock")
         return x
 class PatchEmbed(nn.Module):
     """
@@ -1221,7 +1347,7 @@ class LayerNormChannel(nn.Module):
             + self.bias.unsqueeze(-1).unsqueeze(-1)
         return x
 
-class FeedForwardFNet(nn.Module):
+class FeedForward(nn.Module):
     def __init__(self, dim, hidden_dim, dropout = 0.):
         super().__init__()
         self.net = nn.Sequential(
@@ -1257,7 +1383,7 @@ class FNet(nn.Module):
         for _ in range(depth):
             self.layers.append(nn.ModuleList([
                 PreNorm(dim, FNetBlock()),
-                PreNorm(dim, FeedForwardFNet(dim, mlp_dim, dropout = dropout))
+                PreNorm(dim, FeedForward(dim, mlp_dim, dropout = dropout))
             ]))
     def forward(self, x):
         for attn, ff in self.layers: