| 139 | return x_cls |
| 140 | |
| 141 | class Patch_Attention(nn.Module): |
| 142 | # taken from https://github.com/rwightman/pytorch-image-models/blob/master/timm/models/vision_transformer.py |
| 143 | # with slight modifications to do CA |
| 144 | def __init__(self, dim, num_heads=8, qkv_bias=False, qk_scale=None, attn_drop=0., proj_drop=0.): |
| 145 | super().__init__() |
| 146 | self.num_heads = num_heads |
| 147 | head_dim = dim // num_heads |
| 148 | self.scale = qk_scale or head_dim ** -0.5 |
| 149 | |
| 150 | self.q = nn.Linear(dim, dim, bias=qkv_bias) |
| 151 | self.k = nn.Linear(dim, dim, bias=qkv_bias) |
| 152 | self.v = nn.Linear(dim, dim, bias=qkv_bias) |
| 153 | self.attn_drop = nn.Dropout(attn_drop) |
| 154 | self.proj = nn.Linear(dim, dim) |
| 155 | self.proj_drop = nn.Dropout(proj_drop) |
| 156 | |
| 157 | def forward(self, x): |
| 158 | B, N, C = x.shape |
| 159 | q = self.q(x[:,1:]).unsqueeze(1).reshape(B, N-1, self.num_heads, C // self.num_heads).permute(0, 2, 1, 3) |
| 160 | k = self.k(x).reshape(B, N, self.num_heads, C // self.num_heads).permute(0, 2, 1, 3) |
| 161 | |
| 162 | q = q * self.scale |
| 163 | v = self.v(x).reshape(B, N, self.num_heads, C // self.num_heads).permute(0, 2, 1, 3) |
| 164 | |
| 165 | attn = (q @ k.transpose(-2, -1)) |
| 166 | attn = attn.softmax(dim=-1) |
| 167 | attn = self.attn_drop(attn) |
| 168 | |
| 169 | x_cls = (attn @ v).transpose(1, 2).reshape(B, N-1, C) |
| 170 | x_cls = self.proj(x_cls) |
| 171 | x_cls = self.proj_drop(x_cls) |
| 172 | |
| 173 | return x_cls, attn |
| 174 | |
| 175 | class Patch_Block(nn.Module): |
| 176 | def __init__(self, dim, num_heads, mlp_ratio=4., qkv_bias=False, qk_scale=None, drop=0., attn_drop=0., |