| 90 | """ |
| 91 | |
| 92 | def __init__( |
| 93 | self, |
| 94 | dim, |
| 95 | window_size, |
| 96 | num_heads, |
| 97 | qkv_bias=True, |
| 98 | qk_scale=None, |
| 99 | attn_drop=0.0, |
| 100 | proj_drop=0.0, |
| 101 | ): |
| 102 | |
| 103 | super().__init__() |
| 104 | self.dim = dim |
| 105 | self.window_size = window_size # Wh, Ww |
| 106 | self.num_heads = num_heads |
| 107 | head_dim = dim // num_heads |
| 108 | self.scale = qk_scale or head_dim ** -0.5 |
| 109 | |
| 110 | # define a parameter table of relative position bias |
| 111 | self.relative_position_bias_table = nn.Parameter( |
| 112 | torch.zeros((2 * window_size[0] - 1) * (2 * window_size[1] - 1), num_heads) |
| 113 | ) # 2*Wh-1 * 2*Ww-1, nH |
| 114 | |
| 115 | # get pair-wise relative position index for each token inside the window |
| 116 | coords_h = torch.arange(self.window_size[0]) |
| 117 | coords_w = torch.arange(self.window_size[1]) |
| 118 | coords = torch.stack(torch.meshgrid([coords_h, coords_w])) # 2, Wh, Ww |
| 119 | coords_flatten = torch.flatten(coords, 1) # 2, Wh*Ww |
| 120 | relative_coords = coords_flatten[:, :, None] - coords_flatten[:, None, :] # 2, Wh*Ww, Wh*Ww |
| 121 | relative_coords = relative_coords.permute(1, 2, 0).contiguous() # Wh*Ww, Wh*Ww, 2 |
| 122 | relative_coords[:, :, 0] += self.window_size[0] - 1 # shift to start from 0 |
| 123 | relative_coords[:, :, 1] += self.window_size[1] - 1 |
| 124 | relative_coords[:, :, 0] *= 2 * self.window_size[1] - 1 |
| 125 | relative_position_index = relative_coords.sum(-1) # Wh*Ww, Wh*Ww |
| 126 | self.register_buffer("relative_position_index", relative_position_index) |
| 127 | |
| 128 | self.qkv = nn.Linear(dim, dim * 3, bias=qkv_bias) |
| 129 | self.attn_drop = nn.Dropout(attn_drop) |
| 130 | self.proj = nn.Linear(dim, dim) |
| 131 | self.proj_drop = nn.Dropout(proj_drop) |
| 132 | |
| 133 | trunc_normal_(self.relative_position_bias_table, std=0.02) |
| 134 | self.softmax = nn.Softmax(dim=-1) |
| 135 | |
| 136 | def forward(self, x, mask=None): |
| 137 | """Forward function. |