Open yangchengbang opened 1 month ago
I want to use Encoder as the backbone network to extract features. Does the code have a pretrained model for Encoder?
Hi,may I ask you aquestion about log_fg_matrix = torch.where(ltr, _log_fg_matrix[:, :, 1:, 1:], -float("inf")) # (B, NH, S, S) RuntimeError: expected scalar type float but found double have you met this problem
I want to use Encoder as the backbone network to extract features. Does the code have a pretrained model for Encoder?