1392 lines
54 KiB
Python
1392 lines
54 KiB
Python
import torch
|
||
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 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):
|
||
def __init__(self, params, ):
|
||
super(ConvE, self).__init__()
|
||
|
||
self.p = params
|
||
|
||
self.ent_embed = torch.nn.Embedding(
|
||
self.p.num_ent, self.p.embed_dim, padding_idx=None)
|
||
torch.nn.init.xavier_normal_(self.ent_embed.weight)
|
||
|
||
self.rel_embed = torch.nn.Embedding(
|
||
self.p.num_rel*2, self.p.embed_dim, padding_idx=None)
|
||
torch.nn.init.xavier_normal_(self.rel_embed.weight)
|
||
|
||
self.in_channels = self.p.in_channels
|
||
self.out_channels = self.p.out_channels
|
||
|
||
self.bceloss = torch.nn.BCELoss()
|
||
|
||
self.inp_drop = torch.nn.Dropout(self.p.inp_drop)
|
||
self.hidden_drop = torch.nn.Dropout(self.p.hid_drop)
|
||
self.feature_map_drop = torch.nn.Dropout2d(self.p.feat_drop)
|
||
|
||
self.conv1 = torch.nn.Conv2d(
|
||
self.in_channels, self.out_channels, (self.p.filt_h, self.p.filt_w), 1, 0, bias=True)
|
||
self.bn0 = torch.nn.BatchNorm2d(self.in_channels)
|
||
self.bn1 = torch.nn.BatchNorm2d(self.out_channels)
|
||
self.bn2 = torch.nn.BatchNorm1d(self.p.embed_dim)
|
||
|
||
self.register_parameter(
|
||
'bias', torch.nn.Parameter(torch.zeros(self.p.num_ent)))
|
||
fc_length = (20-self.p.filt_h+1)*(20-self.p.filt_w+1)*self.out_channels
|
||
self.fc = torch.nn.Linear(fc_length, self.p.embed_dim)
|
||
|
||
def loss(self, pred, true_label=None, sub_samp=None):
|
||
label_pos = true_label[0]
|
||
label_neg = true_label[1:]
|
||
loss = self.bceloss(pred, true_label)
|
||
return loss
|
||
|
||
def forward(self, sub, rel, neg_ents, strategy='one_to_x'):
|
||
sub_emb = self.ent_embed(sub).view(-1, 1, self.p.k_w, self.p.k_h)
|
||
rel_emb = self.rel_embed(rel).view(-1, 1, self.p.k_w, self.p.k_h)
|
||
|
||
x = torch.cat([sub_emb, rel_emb], 2)
|
||
x = self.bn0(x)
|
||
x = self.inp_drop(x)
|
||
x = self.conv1(x)
|
||
x = self.bn1(x)
|
||
x = F.relu(x)
|
||
x = self.feature_map_drop(x)
|
||
x = x.view(sub_emb.size(0), -1)
|
||
x = self.fc(x)
|
||
x = self.hidden_drop(x)
|
||
x = self.bn2(x)
|
||
x = F.relu(x)
|
||
|
||
if strategy == 'one_to_n':
|
||
x = torch.mm(x, self.ent_embed.weight.transpose(1, 0))
|
||
x += self.bias.expand_as(x)
|
||
else:
|
||
x = torch.mul(x.unsqueeze(1), self.ent_embed(neg_ents)).sum(dim=-1)
|
||
x += self.bias[neg_ents]
|
||
|
||
pred = torch.sigmoid(x)
|
||
|
||
return pred
|
||
|
||
|
||
class HypER(torch.nn.Module):
|
||
def __init__(self, params, ):
|
||
super(HypER, self).__init__()
|
||
|
||
self.p = params
|
||
|
||
self.ent_embed = torch.nn.Embedding(
|
||
self.p.num_ent, self.p.embed_dim, padding_idx=None)
|
||
torch.nn.init.xavier_normal_(self.ent_embed.weight)
|
||
|
||
self.rel_embed = torch.nn.Embedding(
|
||
self.p.num_rel*2, self.p.embed_dim, padding_idx=None)
|
||
torch.nn.init.xavier_normal_(self.rel_embed.weight)
|
||
|
||
self.in_channels = self.p.in_channels
|
||
self.out_channels = self.p.out_channels
|
||
|
||
self.bceloss = torch.nn.BCELoss()
|
||
|
||
self.inp_drop = torch.nn.Dropout(self.p.inp_drop)
|
||
self.hidden_drop = torch.nn.Dropout(self.p.hid_drop)
|
||
self.feature_map_drop = torch.nn.Dropout2d(self.p.feat_drop)
|
||
|
||
self.bn0 = torch.nn.BatchNorm2d(self.in_channels)
|
||
self.bn1 = torch.nn.BatchNorm2d(self.out_channels)
|
||
self.bn2 = torch.nn.BatchNorm1d(self.p.embed_dim)
|
||
|
||
self.register_parameter(
|
||
'bias', torch.nn.Parameter(torch.zeros(self.p.num_ent)))
|
||
|
||
fc_length = (1-self.p.filt_h+1)*(self.p.embed_dim -
|
||
self.p.filt_w+1)*self.out_channels
|
||
self.fc = torch.nn.Linear(fc_length, self.p.embed_dim)
|
||
fc1_length = self.in_channels*self.out_channels*self.p.filt_h*self.p.filt_w
|
||
self.fc1 = torch.nn.Linear(self.p.embed_dim, fc1_length)
|
||
|
||
def loss(self, pred, true_label=None, sub_samp=None):
|
||
label_pos = true_label[0]
|
||
label_neg = true_label[1:]
|
||
loss = self.bceloss(pred, true_label)
|
||
return loss
|
||
|
||
def forward(self, sub, rel, neg_ents, strategy='one_to_x'):
|
||
sub_emb = self.ent_embed(
|
||
sub).view(-1, 1, 1, self.ent_embed.weight.size(1))
|
||
rel_emb = self.rel_embed(rel)
|
||
|
||
x = self.bn0(sub_emb)
|
||
x = self.inp_drop(x)
|
||
|
||
k = self.fc1(rel_emb)
|
||
k = k.view(-1, self.in_channels, self.out_channels,
|
||
self.p.filt_h, self.p.filt_w)
|
||
k = k.view(sub_emb.size(0)*self.in_channels *
|
||
self.out_channels, 1, self.p.filt_h, self.p.filt_w)
|
||
|
||
x = x.permute(1, 0, 2, 3)
|
||
|
||
x = F.conv2d(x, k, groups=sub_emb.size(0))
|
||
x = x.view(sub_emb.size(0), 1, self.out_channels, 1 -
|
||
self.p.filt_h+1, sub_emb.size(3)-self.p.filt_w+1)
|
||
x = x.permute(0, 3, 4, 1, 2)
|
||
x = torch.sum(x, dim=3)
|
||
x = x.permute(0, 3, 1, 2).contiguous()
|
||
|
||
x = self.bn1(x)
|
||
x = self.feature_map_drop(x)
|
||
x = x.view(sub_emb.size(0), -1)
|
||
x = self.fc(x)
|
||
x = self.hidden_drop(x)
|
||
x = self.bn2(x)
|
||
x = F.relu(x)
|
||
|
||
if strategy == 'one_to_n':
|
||
x = torch.mm(x, self.ent_embed.weight.transpose(1, 0))
|
||
x += self.bias.expand_as(x)
|
||
else:
|
||
x = torch.mul(x.unsqueeze(1), self.ent_embed(neg_ents)).sum(dim=-1)
|
||
x += self.bias[neg_ents]
|
||
|
||
pred = torch.sigmoid(x)
|
||
|
||
return pred
|
||
|
||
|
||
class HypE(torch.nn.Module):
|
||
def __init__(self, params, ):
|
||
super(HypE, self).__init__()
|
||
|
||
self.p = params
|
||
|
||
self.ent_embed = torch.nn.Embedding(
|
||
self.p.num_ent, self.p.embed_dim, padding_idx=None)
|
||
torch.nn.init.xavier_normal_(self.ent_embed.weight)
|
||
|
||
self.rel_embed = torch.nn.Embedding(
|
||
self.p.num_rel*2, self.p.embed_dim, padding_idx=None)
|
||
torch.nn.init.xavier_normal_(self.rel_embed.weight)
|
||
|
||
self.in_channels = self.p.in_channels
|
||
self.out_channels = self.p.out_channels
|
||
|
||
self.bceloss = torch.nn.BCELoss()
|
||
|
||
self.inp_drop = torch.nn.Dropout(self.p.inp_drop)
|
||
self.hidden_drop = torch.nn.Dropout(self.p.hid_drop)
|
||
self.feature_map_drop = torch.nn.Dropout2d(self.p.feat_drop)
|
||
|
||
self.bn0 = torch.nn.BatchNorm2d(self.in_channels)
|
||
self.bn1 = torch.nn.BatchNorm2d(self.out_channels)
|
||
self.bn2 = torch.nn.BatchNorm1d(self.p.embed_dim)
|
||
|
||
self.register_parameter(
|
||
'bias', torch.nn.Parameter(torch.zeros(self.p.num_ent)))
|
||
|
||
fc_length = (10-self.p.filt_h+1)*(20-self.p.filt_w+1)*self.out_channels
|
||
self.fc = torch.nn.Linear(fc_length, self.p.embed_dim)
|
||
|
||
def loss(self, pred, true_label=None, sub_samp=None):
|
||
label_pos = true_label[0]
|
||
label_neg = true_label[1:]
|
||
loss = self.bceloss(pred, true_label)
|
||
return loss
|
||
|
||
def forward(self, sub, rel, neg_ents, strategy='one_to_x'):
|
||
sub_emb = self.ent_embed(
|
||
sub).view(-1, 1, 1, self.ent_embed.weight.size(1))
|
||
rel_emb = self.rel_embed(rel)
|
||
|
||
x = self.bn0(sub_emb)
|
||
x = self.inp_drop(x)
|
||
|
||
k = self.fc1(rel_emb)
|
||
k = k.view(-1, self.in_channels, self.out_channels,
|
||
self.p.filt_h, self.p.filt_w)
|
||
k = k.view(sub_emb.size(0)*self.in_channels *
|
||
self.out_channels, 1, self.p.filt_h, self.p.filt_w)
|
||
|
||
x = x.permute(1, 0, 2, 3)
|
||
|
||
x = F.conv2d(x, k, groups=sub_emb.size(0))
|
||
x = x.view(sub_emb.size(0), 1, self.out_channels, 1 -
|
||
self.p.filt_h+1, sub_emb.size(3)-self.p.filt_w+1)
|
||
x = x.permute(0, 3, 4, 1, 2)
|
||
x = torch.sum(x, dim=3)
|
||
x = x.permute(0, 3, 1, 2).contiguous()
|
||
|
||
x = self.bn1(x)
|
||
x = self.feature_map_drop(x)
|
||
x = x.view(sub_emb.size(0), -1)
|
||
x = self.fc(x)
|
||
x = self.hidden_drop(x)
|
||
x = self.bn2(x)
|
||
x = F.relu(x)
|
||
|
||
if strategy == 'one_to_n':
|
||
x = torch.mm(x, self.ent_embed.weight.transpose(1, 0))
|
||
x += self.bias.expand_as(x)
|
||
else:
|
||
x = torch.mul(x.unsqueeze(1), self.ent_embed(neg_ents)).sum(dim=-1)
|
||
x += self.bias[neg_ents]
|
||
|
||
pred = torch.sigmoid(x)
|
||
|
||
return pred
|
||
|
||
|
||
class DistMult(torch.nn.Module):
|
||
def __init__(self, params, ):
|
||
super(DistMult, self).__init__()
|
||
|
||
self.p = params
|
||
|
||
self.ent_embed = torch.nn.Embedding(
|
||
self.p.num_ent, self.p.embed_dim, padding_idx=None)
|
||
torch.nn.init.xavier_normal_(self.ent_embed.weight)
|
||
|
||
self.rel_embed = torch.nn.Embedding(
|
||
self.p.num_rel*2, self.p.embed_dim, padding_idx=None)
|
||
torch.nn.init.xavier_normal_(self.rel_embed.weight)
|
||
|
||
self.bceloss = torch.nn.BCELoss()
|
||
|
||
self.inp_drop = torch.nn.Dropout(self.p.inp_drop)
|
||
self.bn0 = torch.nn.BatchNorm1d(self.p.embed_dim)
|
||
|
||
self.register_parameter(
|
||
'bias', torch.nn.Parameter(torch.zeros(self.p.num_ent)))
|
||
|
||
def loss(self, pred, true_label=None, sub_samp=None):
|
||
label_pos = true_label[0]
|
||
label_neg = true_label[1:]
|
||
loss = self.bceloss(pred, true_label)
|
||
return loss
|
||
|
||
def forward(self, sub, rel, neg_ents, strategy='one_to_x'):
|
||
sub_emb = self.ent_embed(sub)
|
||
rel_emb = self.rel_embed(rel)
|
||
sub_emb = self.bn0(sub_emb)
|
||
sub_emb = self.inp_drop(sub_emb)
|
||
|
||
if strategy == 'one_to_n':
|
||
x = torch.mm(sub_emb * rel_emb,
|
||
self.ent_embed.weight.transpose(1, 0))
|
||
x += self.bias.expand_as(x)
|
||
else:
|
||
x = torch.mul((sub_emb * rel_emb).unsqueeze(1),
|
||
self.ent_embed(neg_ents)).sum(dim=-1)
|
||
x += self.bias[neg_ents]
|
||
|
||
pred = torch.sigmoid(x)
|
||
|
||
return pred
|
||
|
||
|
||
class ComplEx(torch.nn.Module):
|
||
def __init__(self, params, ):
|
||
super(ComplEx, self).__init__()
|
||
|
||
self.p = params
|
||
|
||
self.ent_embed_real = torch.nn.Embedding(
|
||
self.p.num_ent, self.p.embed_dim, padding_idx=None)
|
||
torch.nn.init.xavier_normal_(self.ent_embed_real.weight)
|
||
|
||
self.ent_embed_imaginary = torch.nn.Embedding(
|
||
self.p.num_ent, self.p.embed_dim, padding_idx=None)
|
||
torch.nn.init.xavier_normal_(self.ent_embed_imaginary.weight)
|
||
|
||
self.rel_embed_real = torch.nn.Embedding(
|
||
self.p.num_rel*2, self.p.embed_dim, padding_idx=None)
|
||
torch.nn.init.xavier_normal_(self.rel_embed_real.weight)
|
||
|
||
self.rel_embed_imaginary = torch.nn.Embedding(
|
||
self.p.num_rel*2, self.p.embed_dim, padding_idx=None)
|
||
torch.nn.init.xavier_normal_(self.rel_embed_imaginary.weight)
|
||
|
||
self.bceloss = torch.nn.BCELoss()
|
||
|
||
self.inp_drop = torch.nn.Dropout(self.p.inp_drop)
|
||
self.bn0 = torch.nn.BatchNorm1d(self.p.embed_dim)
|
||
self.bn1 = torch.nn.BatchNorm1d(self.p.embed_dim)
|
||
|
||
self.register_parameter(
|
||
'bias', torch.nn.Parameter(torch.zeros(self.p.num_ent)))
|
||
|
||
def loss(self, pred, true_label=None, sub_samp=None):
|
||
label_pos = true_label[0]
|
||
label_neg = true_label[1:]
|
||
loss = self.bceloss(pred, true_label)
|
||
return loss
|
||
|
||
def forward(self, sub, rel, neg_ents, strategy='one_to_x'):
|
||
sub_emb_real = self.ent_embed_real(sub)
|
||
sub_emb_imaginary = self.ent_embed_imaginary(sub)
|
||
|
||
rel_emb_real = self.rel_embed_real(rel)
|
||
rel_emb_imaginary = self.rel_embed_imaginary(rel)
|
||
|
||
sub_emb_real = self.bn0(sub_emb_real)
|
||
sub_emb_real = self.inp_drop(sub_emb_real)
|
||
|
||
sub_emb_imaginary = self.bn0(sub_emb_imaginary)
|
||
sub_emb_imaginary = self.inp_drop(sub_emb_imaginary)
|
||
|
||
if strategy == 'one_to_n':
|
||
x = torch.mm(sub_emb_real*rel_emb_real, self.ent_embed_real.weight.transpose(1, 0)) +\
|
||
torch.mm(sub_emb_real*rel_emb_imaginary, self.ent_embed_imaginary.weight.transpose(1, 0)) +\
|
||
torch.mm(sub_emb_imaginary*rel_emb_real, self.ent_embed_imaginary.weight.transpose(1, 0)) -\
|
||
torch.mm(sub_emb_imaginary*rel_emb_imaginary,
|
||
self.ent_embed_real.weight.transpose(1, 0))
|
||
x += self.bias.expand_as(x)
|
||
else:
|
||
neg_embs_real = self.ent_embed_real(neg_ents)
|
||
neg_embs_imaginary = self.ent_embed_imaginary(neg_ents)
|
||
|
||
x = (torch.mul((sub_emb_real*rel_emb_real).unsqueeze(1), neg_embs_real) +
|
||
torch.mul((sub_emb_real*rel_emb_imaginary).unsqueeze(1), neg_embs_imaginary) +
|
||
torch.mul((sub_emb_imaginary*rel_emb_real).unsqueeze(1), neg_embs_imaginary) -
|
||
torch.mul((sub_emb_imaginary*rel_emb_imaginary).unsqueeze(1), neg_embs_real)).sum(dim=-1)
|
||
|
||
x += self.bias[neg_ents]
|
||
|
||
pred = torch.sigmoid(x)
|
||
|
||
return pred
|
||
|
||
|
||
class TuckER(torch.nn.Module):
|
||
def __init__(self, params, ):
|
||
super(TuckER, self).__init__()
|
||
|
||
self.p = params
|
||
|
||
self.ent_embed = torch.nn.Embedding(
|
||
self.p.num_ent, self.p.embed_dim, padding_idx=None)
|
||
torch.nn.init.xavier_normal_(self.ent_embed.weight)
|
||
|
||
self.rel_embed = torch.nn.Embedding(
|
||
self.p.num_rel*2, self.p.embed_dim, padding_idx=None)
|
||
torch.nn.init.xavier_normal_(self.rel_embed.weight)
|
||
|
||
self.core_W = torch.nn.Parameter(torch.tensor(np.random.uniform(-1, 1, (self.p.embed_dim,
|
||
self.p.embed_dim, self.p.embed_dim)), dtype=torch.float, device="cuda", requires_grad=True))
|
||
|
||
self.bceloss = torch.nn.BCELoss()
|
||
|
||
self.inp_drop = torch.nn.Dropout(self.p.inp_drop)
|
||
self.hidden_drop1 = torch.nn.Dropout(self.p.hid_drop)
|
||
self.hidden_drop2 = torch.nn.Dropout(self.p.hid_drop)
|
||
|
||
self.bn0 = torch.nn.BatchNorm1d(self.p.embed_dim)
|
||
self.bn1 = torch.nn.BatchNorm1d(self.p.embed_dim)
|
||
|
||
self.register_parameter(
|
||
'bias', torch.nn.Parameter(torch.zeros(self.p.num_ent)))
|
||
|
||
def loss(self, pred, true_label=None, sub_samp=None):
|
||
label_pos = true_label[0]
|
||
label_neg = true_label[1:]
|
||
loss = self.bceloss(pred, true_label)
|
||
return loss
|
||
|
||
def forward(self, sub, rel, neg_ents, strategy='one_to_x'):
|
||
sub_emb = self.ent_embed(sub)
|
||
|
||
sub_emb = self.bn0(sub_emb)
|
||
sub_emb = self.inp_drop(sub_emb)
|
||
|
||
x = sub_emb.view(-1, 1, sub_emb.size(1))
|
||
|
||
r = self.rel_embed(rel)
|
||
|
||
W_mat = torch.mm(r, self.core_W.view(r.size(1), -1))
|
||
W_mat = W_mat.view(-1, sub_emb.size(1), sub_emb.size(1))
|
||
W_mat = self.hidden_drop1(W_mat)
|
||
|
||
x = torch.bmm(x, W_mat)
|
||
x = x.view(-1, sub_emb.size(1))
|
||
x = self.bn1(x)
|
||
x = self.hidden_drop2(x)
|
||
|
||
if strategy == 'one_to_n':
|
||
x = torch.mm(x, self.ent_embed.weight.transpose(1, 0))
|
||
x += self.bias.expand_as(x)
|
||
else:
|
||
x = torch.mul(x.unsqueeze(1),
|
||
self.ent_embed(neg_ents)).sum(dim=-1)
|
||
x += self.bias[neg_ents]
|
||
|
||
pred = torch.sigmoid(x)
|
||
|
||
return pred
|
||
|
||
|
||
class FouriER(torch.nn.Module):
|
||
def __init__(self, params, hid_drop = None, embed_dim = None):
|
||
super(FouriER, self).__init__()
|
||
|
||
if hid_drop is not None:
|
||
self.p.hid_drop = hid_drop
|
||
if embed_dim is not None:
|
||
self.p.ent_vec_dim = embed_dim
|
||
self.p.rel_vec_dim = embed_dim
|
||
self.p.embed_dim = embed_dim
|
||
|
||
self.p = params
|
||
|
||
image_h, image_w = self.p.image_h, self.p.image_w
|
||
|
||
self.in_channels = self.p.in_channels
|
||
self.out_channels = self.p.out_channels
|
||
|
||
self.ent_embed = torch.nn.Embedding(
|
||
self.p.num_ent, self.p.ent_vec_dim, padding_idx=None)
|
||
torch.nn.init.xavier_normal_(self.ent_embed.weight)
|
||
|
||
self.rel_embed = torch.nn.Embedding(
|
||
self.p.num_rel*2, self.p.rel_vec_dim, padding_idx=None)
|
||
torch.nn.init.xavier_normal_(self.rel_embed.weight)
|
||
|
||
self.ent_fusion = torch.nn.Linear(
|
||
self.p.ent_vec_dim, image_h*image_w)
|
||
torch.nn.init.xavier_normal_(self.ent_fusion.weight)
|
||
|
||
self.rel_fusion = torch.nn.Linear(
|
||
self.p.rel_vec_dim, image_h*image_w)
|
||
torch.nn.init.xavier_normal_(self.rel_fusion.weight)
|
||
|
||
self.bceloss = torch.nn.BCELoss()
|
||
|
||
channels = 2
|
||
self.bn0 = torch.nn.BatchNorm2d(channels)
|
||
self.bn1 = torch.nn.BatchNorm1d(self.p.embed_dim)
|
||
self.hidden_drop = torch.nn.Dropout(self.p.hid_drop)
|
||
|
||
self.register_parameter(
|
||
'bias', torch.nn.Parameter(torch.zeros(self.p.num_ent)))
|
||
|
||
patch_size = self.p.patch_size
|
||
|
||
assert (image_h % patch_size) == 0 and (image_w %
|
||
patch_size) == 0, 'image must be divisible by patch size'
|
||
|
||
self.patch_embed = PatchEmbed(in_chans=channels, patch_size=self.p.patch_size,
|
||
embed_dim=self.p.embed_dim, stride=4, padding=2)
|
||
network = []
|
||
layers = [4, 4, 12, 4]
|
||
embed_dims = [self.p.embed_dim, 128, 320, 128]
|
||
mlp_ratios = [4, 4, 4, 4]
|
||
downsamples = [True, True, True, True]
|
||
pool_size=3
|
||
act_layer=nn.GELU
|
||
drop_rate=self.p.drop
|
||
norm_layer=GroupNorm
|
||
drop_path_rate=self.p.drop_path
|
||
use_layer_scale=True
|
||
layer_scale_init_value=1e-5
|
||
down_patch_size=3
|
||
down_stride=2
|
||
down_pad=1
|
||
num_classes=self.p.embed_dim
|
||
for i in range(len(layers)):
|
||
stage = basic_blocks(embed_dims[i], i, layers,
|
||
pool_size=pool_size, mlp_ratio=mlp_ratios[i],
|
||
act_layer=act_layer, norm_layer=norm_layer,
|
||
drop_rate=drop_rate,
|
||
drop_path_rate=drop_path_rate,
|
||
use_layer_scale=use_layer_scale,
|
||
layer_scale_init_value=layer_scale_init_value)
|
||
network.append(stage)
|
||
if i >= len(layers) - 1:
|
||
break
|
||
if downsamples[i] or embed_dims[i] != embed_dims[i+1]:
|
||
# downsampling between two stages
|
||
network.append(
|
||
PatchEmbed(
|
||
patch_size=down_patch_size, stride=down_stride,
|
||
padding=down_pad,
|
||
in_chans=embed_dims[i], embed_dim=embed_dims[i+1]
|
||
)
|
||
)
|
||
|
||
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()
|
||
|
||
|
||
def loss(self, pred, true_label=None, sub_samp=None):
|
||
label_pos = true_label[0]
|
||
label_neg = true_label[1:]
|
||
loss = self.bceloss(pred, true_label)
|
||
return loss
|
||
|
||
def forward_embeddings(self, x):
|
||
x = self.patch_embed(x)
|
||
return x
|
||
|
||
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):
|
||
try:
|
||
x = block(x, graph)
|
||
except:
|
||
x = block(x)
|
||
# output only the features of last layer for image classification
|
||
return x
|
||
|
||
def forward(self, sub, rel, neg_ents, strategy='one_to_x'):
|
||
sub_emb = self.ent_fusion(self.ent_embed(sub))
|
||
rel_emb = self.rel_fusion(self.rel_embed(rel))
|
||
comb_emb = torch.stack([sub_emb.view(-1, self.p.image_h, self.p.image_w), rel_emb.view(-1, self.p.image_h, self.p.image_w)], dim=1)
|
||
y = comb_emb.view(-1, 2, self.p.image_h, self.p.image_w)
|
||
y = self.bn0(y)
|
||
z = self.forward_embeddings(y)
|
||
z = self.forward_tokens(z)
|
||
z = z.mean([-2, -1])
|
||
z = self.norm(z)
|
||
x = self.head(z)
|
||
x = self.hidden_drop(x)
|
||
x = self.bn1(x)
|
||
x = F.relu(x)
|
||
|
||
if strategy == 'one_to_n':
|
||
x = torch.mm(x, self.ent_embed.weight.transpose(1, 0))
|
||
x += self.bias.expand_as(x)
|
||
else:
|
||
x = torch.mul(x.unsqueeze(1),
|
||
self.ent_embed(neg_ents)).sum(dim=-1)
|
||
x += self.bias[neg_ents]
|
||
|
||
pred = torch.sigmoid(x)
|
||
|
||
return pred
|
||
|
||
|
||
class InteractE(torch.nn.Module):
|
||
"""
|
||
Proposed method in the paper. Refer Section 6 of the paper for mode details
|
||
Parameters
|
||
----------
|
||
params: Hyperparameters of the model
|
||
chequer_perm: Reshaping to be used by the model
|
||
|
||
Returns
|
||
-------
|
||
The InteractE model instance
|
||
|
||
"""
|
||
|
||
def __init__(self, params, chequer_perm):
|
||
super(InteractE, self).__init__()
|
||
|
||
self.p = params
|
||
|
||
self.ent_embed = torch.nn.Embedding(
|
||
self.p.num_ent, self.p.embed_dim, padding_idx=None)
|
||
torch.nn.init.xavier_normal_(self.ent_embed.weight)
|
||
|
||
self.rel_embed = torch.nn.Embedding(
|
||
self.p.num_rel*2, self.p.embed_dim, padding_idx=None)
|
||
torch.nn.init.xavier_normal_(self.rel_embed.weight)
|
||
|
||
self.bceloss = torch.nn.BCELoss()
|
||
|
||
self.hidden_drop = torch.nn.Dropout(self.p.hid_drop)
|
||
self.feature_map_drop = torch.nn.Dropout2d(self.p.feat_drop)
|
||
self.bn0 = torch.nn.BatchNorm2d(self.p.perm)
|
||
|
||
flat_sz_h = self.p.k_h
|
||
flat_sz_w = 2*self.p.k_w
|
||
self.padding = 0
|
||
|
||
self.bn1 = torch.nn.BatchNorm2d(self.p.num_filt*self.p.perm)
|
||
self.flat_sz = flat_sz_h * flat_sz_w * self.p.num_filt*self.p.perm
|
||
|
||
self.bn2 = torch.nn.BatchNorm1d(self.p.embed_dim)
|
||
self.fc = torch.nn.Linear(self.flat_sz, self.p.embed_dim)
|
||
self.chequer_perm = chequer_perm
|
||
|
||
self.register_parameter(
|
||
'bias', torch.nn.Parameter(torch.zeros(self.p.num_ent)))
|
||
self.register_parameter('conv_filt', torch.nn.Parameter(
|
||
torch.zeros(self.p.num_filt, 1, self.p.ker_sz, self.p.ker_sz)))
|
||
|
||
torch.nn.init.xavier_normal_(self.conv_filt)
|
||
|
||
def loss(self, pred, true_label=None, sub_samp=None):
|
||
label_pos = true_label[0]
|
||
label_neg = true_label[1:]
|
||
loss = self.bceloss(pred, true_label)
|
||
return loss
|
||
|
||
def circular_padding_chw(self, batch, padding):
|
||
upper_pad = batch[..., -padding:, :]
|
||
lower_pad = batch[..., :padding, :]
|
||
temp = torch.cat([upper_pad, batch, lower_pad], dim=2)
|
||
|
||
left_pad = temp[..., -padding:]
|
||
right_pad = temp[..., :padding]
|
||
padded = torch.cat([left_pad, temp, right_pad], dim=3)
|
||
return padded
|
||
|
||
def forward(self, sub, rel, neg_ents, strategy='one_to_x'):
|
||
sub_emb = self.ent_embed(sub)
|
||
rel_emb = self.rel_embed(rel)
|
||
comb_emb = torch.cat([sub_emb, rel_emb], dim=1)
|
||
chequer_perm = comb_emb[:, self.chequer_perm]
|
||
stack_inp = chequer_perm.reshape(
|
||
(-1, self.p.perm, 2*self.p.k_w, self.p.k_h))
|
||
stack_inp = self.bn0(stack_inp)
|
||
x = self.inp_drop(stack_inp)
|
||
x = self.circular_padding_chw(x, self.p.ker_sz//2)
|
||
x = F.conv2d(x, self.conv_filt.repeat(self.p.perm, 1, 1, 1),
|
||
padding=self.padding, groups=self.p.perm)
|
||
x = self.bn1(x)
|
||
x = F.relu(x)
|
||
x = self.feature_map_drop(x)
|
||
x = x.view(-1, self.flat_sz)
|
||
x = self.fc(x)
|
||
x = self.hidden_drop(x)
|
||
x = self.bn2(x)
|
||
x = F.relu(x)
|
||
|
||
if strategy == 'one_to_n':
|
||
x = torch.mm(x, self.ent_embed.weight.transpose(1, 0))
|
||
x += self.bias.expand_as(x)
|
||
else:
|
||
x = torch.mul(x.unsqueeze(1), self.ent_embed(neg_ents)).sum(dim=-1)
|
||
x += self.bias[neg_ents]
|
||
|
||
pred = torch.sigmoid(x)
|
||
|
||
return pred
|
||
|
||
class GroupNorm(nn.GroupNorm):
|
||
"""
|
||
Group Normalization with 1 group.
|
||
Input: tensor in shape [B, C, H, W]
|
||
"""
|
||
def __init__(self, num_channels, **kwargs):
|
||
super().__init__(1, num_channels, **kwargs)
|
||
|
||
def basic_blocks(dim, index, layers,
|
||
pool_size=3, mlp_ratio=4.,
|
||
act_layer=nn.GELU, norm_layer=GroupNorm,
|
||
drop_rate=.0, drop_path_rate=0.,
|
||
use_layer_scale=True, layer_scale_init_value=1e-5):
|
||
"""
|
||
generate PoolFormer blocks for a stage
|
||
return: PoolFormer blocks
|
||
"""
|
||
blocks = []
|
||
for block_idx in range(layers[index]):
|
||
block_dpr = drop_path_rate * (
|
||
block_idx + sum(layers[:index])) / (sum(layers) - 1)
|
||
blocks.append(PoolFormerBlock(
|
||
dim, pool_size=pool_size, mlp_ratio=mlp_ratio,
|
||
act_layer=act_layer, norm_layer=norm_layer,
|
||
drop=drop_rate, drop_path=block_dpr,
|
||
use_layer_scale=use_layer_scale,
|
||
layer_scale_init_value=layer_scale_init_value,
|
||
))
|
||
blocks = SeqModel(*blocks)
|
||
|
||
return blocks
|
||
|
||
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, C, H, W = 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
|
||
|
||
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]):
|
||
|
||
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)
|
||
|
||
# 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.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)
|
||
|
||
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)
|
||
|
||
# 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)).cuda()).exp()
|
||
attn = attn * logit_scale
|
||
|
||
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 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).transpose(1, 2).reshape(B_, N, C)
|
||
x = self.proj(x)
|
||
x = self.proj_drop(x)
|
||
return x
|
||
|
||
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
|
||
|
||
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, -1, H, W)
|
||
return x
|
||
|
||
class SeqModel(nn.Sequential):
|
||
def forward(self, *inputs):
|
||
for module in self._modules.values():
|
||
if type(inputs) == tuple:
|
||
inputs = module(*inputs)
|
||
else:
|
||
inputs = module(inputs)
|
||
return inputs
|
||
|
||
def propagate(x: torch.Tensor, weight: torch.Tensor,
|
||
index_kept: torch.Tensor, index_prop: torch.Tensor,
|
||
standard: str = "None", alpha: Optional[float] = 0,
|
||
token_scales: Optional[torch.Tensor] = None,
|
||
cls_token=True):
|
||
"""
|
||
Propagate tokens based on the selection results.
|
||
================================================
|
||
Args:
|
||
- x: Tensor([B, N, C]): the feature map of N tokens, including the [CLS] token.
|
||
|
||
- weight: Tensor([B, N-1, N-1]): the weight of each token propagated to the other tokens,
|
||
excluding the [CLS] token. weight could be a pre-defined
|
||
graph of the current feature map (by default) or the
|
||
attention map (need to manually modify the Block Module).
|
||
|
||
- index_kept: Tensor([B, N-1-num_prop]): the index of kept image tokens in the feature map X
|
||
|
||
- index_prop: Tensor([B, num_prop]): the index of propagated image tokens in the feature map X
|
||
|
||
- standard: str: the method applied to propagate the tokens, including "None", "Mean" and
|
||
"GraphProp"
|
||
|
||
- alpha: float: the coefficient of propagated features
|
||
|
||
- token_scales: Tensor([B, N]): the scale of tokens, including the [CLS] token. token_scales
|
||
is None by default. If it is not None, then token_scales
|
||
represents the scales of each token and should sum up to N.
|
||
|
||
Return:
|
||
- x: Tensor([B, N-1-num_prop, C]): the feature map after propagation
|
||
|
||
- weight: Tensor([B, N-1-num_prop, N-1-num_prop]): the graph of feature map after propagation
|
||
|
||
- token_scales: Tensor([B, N-1-num_prop]): the scale of tokens after propagation
|
||
"""
|
||
|
||
B, N, C = x.shape
|
||
|
||
# Step 1: divide tokens
|
||
if cls_token:
|
||
x_cls = x[:, 0:1] # B, 1, C
|
||
x_kept = x.gather(dim=1, index=index_kept.unsqueeze(-1).expand(-1,-1,C)) # B, N-1-num_prop, C
|
||
x_prop = x.gather(dim=1, index=index_prop.unsqueeze(-1).expand(-1,-1,C)) # B, num_prop, C
|
||
|
||
# Step 2: divide token_scales if it is not None
|
||
if token_scales is not None:
|
||
if cls_token:
|
||
token_scales_cls = token_scales[:, 0:1] # B, 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
|
||
|
||
# Step 3: propagate tokens
|
||
if standard == "None":
|
||
"""
|
||
No further propagation
|
||
"""
|
||
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
|
||
|
||
|
||
|
||
def select(weight: torch.Tensor, standard: str = "None", num_prop: int = 0, cls_token = True):
|
||
"""
|
||
Select image tokens to be propagated. The [CLS] token will be ignored.
|
||
======================================================================
|
||
Args:
|
||
- weight: Tensor([B, H, N, N]): used for selecting the kept tokens. Only support the
|
||
attention map of tokens at the moment.
|
||
|
||
- standard: str: the method applied to select the tokens
|
||
|
||
- num_prop: int: the number of tokens to be propagated
|
||
|
||
Return:
|
||
- index_kept: Tensor([B, N-1-num_prop]): the index of kept tokens
|
||
|
||
- index_prop: Tensor([B, num_prop]): the index of propagated tokens
|
||
"""
|
||
|
||
assert len(weight.shape) == 4, "Selection methods on tensors other than the attention map haven't been supported yet."
|
||
B, H, N1, N2 = weight.shape
|
||
assert N1 == N2, "Selection methods on tensors other than the attention map haven't been supported yet."
|
||
N = N1
|
||
assert num_prop >= 0, "The number of propagated/pruned tokens must be non-negative."
|
||
|
||
if cls_token:
|
||
if standard == "CLSAttnMean":
|
||
token_rank = weight[:,:,0,1:].mean(1)
|
||
|
||
elif standard == "CLSAttnMax":
|
||
token_rank = weight[:,:,0,1:].max(1)[0]
|
||
|
||
elif standard == "IMGAttnMean":
|
||
token_rank = weight[:,:,:,1:].sum(-2).mean(1)
|
||
|
||
elif standard == "IMGAttnMax":
|
||
token_rank = weight[:,:,:,1:].sum(-2).max(1)[0]
|
||
|
||
elif standard == "DiagAttnMean":
|
||
token_rank = torch.diagonal(weight, dim1=-2, dim2=-1)[:,:,1:].mean(1)
|
||
|
||
elif standard == "DiagAttnMax":
|
||
token_rank = torch.diagonal(weight, dim1=-2, dim2=-1)[:,:,1:].max(1)[0]
|
||
|
||
elif standard == "MixedAttnMean":
|
||
token_rank_1 = torch.diagonal(weight, dim1=-2, dim2=-1)[:,:,1:].mean(1)
|
||
token_rank_2 = weight[:,:,:,1:].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)[:,:,1:].max(1)[0]
|
||
token_rank_2 = weight[:,:,:,1:].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)[:,:,1:].max(1)[0]
|
||
token_rank_2 = weight[:,:,:,1:].sum(-2).max(1)[0]
|
||
token_rank = token_rank_1 + token_rank_2
|
||
|
||
elif standard == "CosSimMean":
|
||
weight = weight[:,:,1:,:].mean(1)
|
||
weight = weight / weight.norm(dim=-1, keepdim=True)
|
||
token_rank = -(weight @ weight.transpose(-1, -2)).sum(-1)
|
||
|
||
elif standard == "CosSimMax":
|
||
weight = weight[:,:,1:,:].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]+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):
|
||
"""
|
||
Implementation of one PoolFormer block.
|
||
--dim: embedding dim
|
||
--pool_size: pooling size
|
||
--mlp_ratio: mlp expansion ratio
|
||
--act_layer: activation
|
||
--norm_layer: normalization
|
||
--drop: dropout rate
|
||
--drop path: Stochastic Depth,
|
||
refer to https://arxiv.org/abs/1603.09382
|
||
--use_layer_scale, --layer_scale_init_value: LayerScale,
|
||
refer to https://arxiv.org/abs/2103.17239
|
||
"""
|
||
def __init__(self, dim, pool_size=3, mlp_ratio=4.,
|
||
act_layer=nn.GELU, norm_layer=GroupNorm,
|
||
drop=0., drop_path=0.,
|
||
use_layer_scale=True, layer_scale_init_value=1e-5):
|
||
|
||
super().__init__()
|
||
|
||
self.norm1 = norm_layer(dim)
|
||
#self.token_mixer = Pooling(pool_size=pool_size)
|
||
# self.token_mixer = FNetBlock()
|
||
self.window_size = 4
|
||
self.attn_mask = None
|
||
self.token_mixer = WindowAttention(dim=dim, window_size=to_2tuple(self.window_size), num_heads=4)
|
||
self.norm2 = norm_layer(dim)
|
||
mlp_hidden_dim = int(dim * mlp_ratio)
|
||
self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim,
|
||
act_layer=act_layer, drop=drop)
|
||
|
||
# The following two techniques are useful to train deep PoolFormers.
|
||
self.drop_path = DropPath(drop_path) if drop_path > 0. \
|
||
else nn.Identity()
|
||
self.use_layer_scale = use_layer_scale
|
||
if use_layer_scale:
|
||
self.layer_scale_1 = nn.Parameter(
|
||
layer_scale_init_value * torch.ones((dim)), requires_grad=True)
|
||
self.layer_scale_2 = nn.Parameter(
|
||
layer_scale_init_value * torch.ones((dim)), requires_grad=True)
|
||
|
||
def forward(self, x, weight, token_scales = None):
|
||
B, C, H, W = x.shape
|
||
x_windows = window_partition(x, self.window_size)
|
||
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 = attn_windows.view(-1, self.window_size, self.window_size, C)
|
||
x_attn = window_reverse(attn_windows, self.window_size, H, W)
|
||
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)
|
||
* x_attn)
|
||
x = x + self.drop_path(
|
||
self.layer_scale_2.unsqueeze(-1).unsqueeze(-1)
|
||
* self.mlp(self.norm2(x)))
|
||
else:
|
||
x = x + self.drop_path(x_attn)
|
||
x = x + self.drop_path(self.mlp(self.norm2(x)))
|
||
return x
|
||
class PatchEmbed(nn.Module):
|
||
"""
|
||
Patch Embedding that is implemented by a layer of conv.
|
||
Input: tensor in shape [B, C, H, W]
|
||
Output: tensor in shape [B, C, H/stride, W/stride]
|
||
"""
|
||
def __init__(self, patch_size=16, stride=16, padding=0,
|
||
in_chans=3, embed_dim=768, norm_layer=None):
|
||
super().__init__()
|
||
patch_size = to_2tuple(patch_size)
|
||
stride = to_2tuple(stride)
|
||
padding = to_2tuple(padding)
|
||
self.proj = nn.Conv2d(in_chans, embed_dim, kernel_size=patch_size,
|
||
stride=stride, padding=padding)
|
||
self.norm = norm_layer(embed_dim) if norm_layer else nn.Identity()
|
||
|
||
def forward(self, x):
|
||
x = self.proj(x)
|
||
x = self.norm(x)
|
||
return x
|
||
|
||
class Pooling(nn.Module):
|
||
"""
|
||
Implementation of pooling for PoolFormer
|
||
--pool_size: pooling size
|
||
"""
|
||
def __init__(self, pool_size=3):
|
||
super().__init__()
|
||
self.pool = nn.AvgPool2d(
|
||
pool_size, stride=1, padding=pool_size//2, count_include_pad=False)
|
||
|
||
def forward(self, x):
|
||
return self.pool(x) - x
|
||
|
||
|
||
class Mlp(nn.Module):
|
||
"""
|
||
Implementation of MLP with 1*1 convolutions.
|
||
Input: tensor with shape [B, C, H, W]
|
||
"""
|
||
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.Conv2d(in_features, hidden_features, 1)
|
||
self.act = act_layer()
|
||
self.fc2 = nn.Conv2d(hidden_features, out_features, 1)
|
||
self.drop = nn.Dropout(drop)
|
||
self.apply(self._init_weights)
|
||
|
||
def _init_weights(self, m):
|
||
if isinstance(m, nn.Conv2d):
|
||
trunc_normal_(m.weight, std=.02)
|
||
if m.bias is not None:
|
||
nn.init.constant_(m.bias, 0)
|
||
|
||
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
|
||
|
||
class LayerNormChannel(nn.Module):
|
||
"""
|
||
LayerNorm only for Channel Dimension.
|
||
Input: tensor in shape [B, C, H, W]
|
||
"""
|
||
def __init__(self, num_channels, eps=1e-05):
|
||
super().__init__()
|
||
self.weight = nn.Parameter(torch.ones(num_channels))
|
||
self.bias = nn.Parameter(torch.zeros(num_channels))
|
||
self.eps = eps
|
||
|
||
def forward(self, x):
|
||
u = x.mean(1, keepdim=True)
|
||
s = (x - u).pow(2).mean(1, keepdim=True)
|
||
x = (x - u) / torch.sqrt(s + self.eps)
|
||
x = self.weight.unsqueeze(-1).unsqueeze(-1) * x \
|
||
+ self.bias.unsqueeze(-1).unsqueeze(-1)
|
||
return x
|
||
|
||
class FeedForward(nn.Module):
|
||
def __init__(self, dim, hidden_dim, dropout = 0.):
|
||
super().__init__()
|
||
self.net = nn.Sequential(
|
||
nn.Linear(dim, hidden_dim),
|
||
nn.GELU(),
|
||
nn.Dropout(dropout),
|
||
nn.Linear(hidden_dim, dim),
|
||
nn.Dropout(dropout)
|
||
)
|
||
def forward(self, x):
|
||
return self.net(x)
|
||
|
||
class PreNorm(nn.Module):
|
||
def __init__(self, dim, fn):
|
||
super().__init__()
|
||
self.norm = nn.LayerNorm(dim)
|
||
self.fn = fn
|
||
def forward(self, x, **kwargs):
|
||
return self.fn(self.norm(x), **kwargs)
|
||
|
||
class FNetBlock(nn.Module):
|
||
def __init__(self):
|
||
super().__init__()
|
||
|
||
def forward(self, x):
|
||
x = torch.fft.fft(torch.fft.fft(x, dim=-1), dim=-2).real
|
||
return x
|
||
|
||
class FNet(nn.Module):
|
||
def __init__(self, dim, depth, mlp_dim, dropout = 0.):
|
||
super().__init__()
|
||
self.layers = nn.ModuleList([])
|
||
for _ in range(depth):
|
||
self.layers.append(nn.ModuleList([
|
||
PreNorm(dim, FNetBlock()),
|
||
PreNorm(dim, FeedForward(dim, mlp_dim, dropout = dropout))
|
||
]))
|
||
def forward(self, x):
|
||
for attn, ff in self.layers:
|
||
x = attn(x) + x
|
||
x = ff(x) + x
|
||
return x |