xdecoder/Instruct-X-Decoder

1# Copyright (c) Facebook, Inc. and its affiliates.2## Modified by Bowen Cheng from: https://github.com/facebookresearch/detr/blob/master/models/position_encoding.py3"""4Various positional encodings for the transformer.5"""6import math7 8import torch9from torch import nn10 11 12class PositionEmbeddingSine(nn.Module):13    """14    This is a more standard version of the position embedding, very similar to the one15    used by the Attention is all you need paper, generalized to work on images.16    """17 18    def __init__(self, num_pos_feats=64, temperature=10000, normalize=False, scale=None):19        super().__init__()20        self.num_pos_feats = num_pos_feats21        self.temperature = temperature22        self.normalize = normalize23        if scale is not None and normalize is False:24            raise ValueError("normalize should be True if scale is passed")25        if scale is None:26            scale = 2 * math.pi27        self.scale = scale28 29    def forward(self, x, mask=None):30        if mask is None:31            mask = torch.zeros((x.size(0), x.size(2), x.size(3)), device=x.device, dtype=torch.bool)32        not_mask = ~mask33        y_embed = not_mask.cumsum(1, dtype=x.dtype)34        x_embed = not_mask.cumsum(2, dtype=x.dtype)35        if self.normalize:36            eps = 1e-637            y_embed = y_embed / (y_embed[:, -1:, :] + eps) * self.scale38            x_embed = x_embed / (x_embed[:, :, -1:] + eps) * self.scale39 40        dim_t = torch.arange(self.num_pos_feats, dtype=x.dtype, device=x.device)41        dim_t = self.temperature ** (2 * (dim_t // 2) / self.num_pos_feats)42 43        pos_x = x_embed[:, :, :, None] / dim_t44        pos_y = y_embed[:, :, :, None] / dim_t45        pos_x = torch.stack(46            (pos_x[:, :, :, 0::2].sin(), pos_x[:, :, :, 1::2].cos()), dim=447        ).flatten(3)48        pos_y = torch.stack(49            (pos_y[:, :, :, 0::2].sin(), pos_y[:, :, :, 1::2].cos()), dim=450        ).flatten(3)51        pos = torch.cat((pos_y, pos_x), dim=3).permute(0, 3, 1, 2)52        return pos53    54    def __repr__(self, _repr_indent=4):55        head = "Positional encoding " + self.__class__.__name__56        body = [57            "num_pos_feats: {}".format(self.num_pos_feats),58            "temperature: {}".format(self.temperature),59            "normalize: {}".format(self.normalize),60            "scale: {}".format(self.scale),61        ]62        # _repr_indent = 463        lines = [head] + [" " * _repr_indent + line for line in body]64        return "\n".join(lines)65