diffusers_repo/examples/research_projects/anytext/ocr_recog/RecCTCHead.py

from torch import nn


class CTCHead(nn.Module):
    def __init__(
        self, in_channels, out_channels=6625, fc_decay=0.0004, mid_channels=None, return_feats=False, **kwargs
    ):
        super(CTCHead, self).__init__()
        if mid_channels is None:
            self.fc = nn.Linear(
                in_channels,
                out_channels,
                bias=True,
            )
        else:
            self.fc1 = nn.Linear(
                in_channels,
                mid_channels,
                bias=True,
            )
            self.fc2 = nn.Linear(
                mid_channels,
                out_channels,
                bias=True,
            )

        self.out_channels = out_channels
        self.mid_channels = mid_channels
        self.return_feats = return_feats

    def forward(self, x, labels=None):
        if self.mid_channels is None:
            predicts = self.fc(x)
        else:
            x = self.fc1(x)
            predicts = self.fc2(x)

        if self.return_feats:
            result = {}
            result["ctc"] = predicts
            result["ctc_neck"] = x
        else:
            result = predicts

        return result