From 1ca8b0b9e0613c1e02f6a5d8b49e20c4d6916412 Mon Sep 17 00:00:00 2001 From: Gustaf Rydholm Date: Thu, 22 Apr 2021 08:15:58 +0200 Subject: Fixed training script, able to train vqvae --- text_recognizer/networks/vqvae/decoder.py | 20 ++++++++++++++++---- 1 file changed, 16 insertions(+), 4 deletions(-) (limited to 'text_recognizer/networks/vqvae/decoder.py') diff --git a/text_recognizer/networks/vqvae/decoder.py b/text_recognizer/networks/vqvae/decoder.py index 8847aba..93a1e43 100644 --- a/text_recognizer/networks/vqvae/decoder.py +++ b/text_recognizer/networks/vqvae/decoder.py @@ -44,7 +44,12 @@ class Decoder(nn.Module): # Configure encoder. self.decoder = self._build_decoder( - channels, kernel_sizes, strides, num_residual_layers, activation, dropout, + channels, + kernel_sizes, + strides, + num_residual_layers, + activation, + dropout, ) def _build_decompression_block( @@ -72,8 +77,10 @@ class Decoder(nn.Module): ) ) - if i < len(self.upsampling): - modules.append(nn.Upsample(size=self.upsampling[i]),) + if self.upsampling and i < len(self.upsampling): + modules.append( + nn.Upsample(size=self.upsampling[i]), + ) if dropout is not None: modules.append(dropout) @@ -102,7 +109,12 @@ class Decoder(nn.Module): ) -> nn.Sequential: self.res_block.append( - nn.Conv2d(self.embedding_dim, channels[0], kernel_size=1, stride=1,) + nn.Conv2d( + self.embedding_dim, + channels[0], + kernel_size=1, + stride=1, + ) ) # Bottleneck module. -- cgit v1.2.3-70-g09d2