Change any output errors

config/celebahq.yaml

@@ -1,5 +1,5 @@
 dataset_config:
-  im_path: '~/HuggingFace/Latent-Diffusion-Conditional/dataset/dataset'
+  im_path: '/home/Latent-Diffusion-Conditional/dataset/dataset'
   im_channels : 3
   im_size : 256
   name: 'celebhq'

dataset/celeba.py

@@ -18,7 +18,7 @@ class ParquetImageDataset(Dataset):
         return len(self.data)
 
     def __getitem__(self, idx):
-        image = Image.open(io.BytesIO(self.data.iloc[idx]['image']))
+        image = Image.open(self.data.iloc[idx]['image'])
         caption = self.data.iloc[idx]['text']
         im_tensor = transforms.Compose([
             transforms.Resize(self.im_size),

models/blocks.py

@@ -101,8 +101,8 @@ class DownBlock(nn.Module):
             ]
         )
 
-        self.resnet_down_conv = nn.Conv2d(in_channels=out_channels, out_channels=out_channels, 4, 2, 1) if self.down_sample else nn.Identity()
-
+        self.resnet_down_conv = nn.Conv2d(out_channels, out_channels,
+                                          4, 2, 1) if self.down_sample else nn.Identity()
     def forward(self, x, t_emb=None, context=None):
         out = x
         for i in range(self.num_layers):

models/vqvae.py

@@ -74,7 +74,7 @@ class VQVAE(nn.Module):
             self.decoder_layers.append(UpBlock(self.down_channels[i], self.down_channels[i-1],
                                        t_emb_dim=None, up_sample=self.down_sample[i-1], num_heads=self.num_heads,
                                        num_layers=self.num_up_layers,
-                                       attn=self.attn[i-1],
+                                       attn=self.attns[i-1],
                                        norm_channels=self.norm_channels))
 
         self.decoder_norm_out = nn.GroupNorm(

scripts/train_vqvae.py → train_vqvae.py

@@ -27,7 +27,7 @@ def train(args):
 
     autoencoder_config = config["autoencoder_params"]
     train_config = config["train_config"]
-    dataset_config = confg["dataset_config"]
+    dataset_config = config["dataset_config"]
 
     # Set seed for reproducability
     seed = train_config["seed"]
@@ -37,7 +37,7 @@ def train(args):
     model = VQVAE(
         im_channels=dataset_config["im_channels"], model_config=autoencoder_config)
 
-    data_loader = create_dataloader(dataset_config[")
+    data_loader = create_dataloader(dataset_config["im_path"])
 
     if not os.path.exists(train_config["task_name"]):
         os.mkdir(train_config["task_name"])
@@ -169,6 +169,6 @@ if __name__ == "__main__":
     parser = argparse.ArgumentParser(
         description="Arguments for vq vae training")
     parser.add_argument("--config_path", type=str,
-                        dest="config_path", default="configs/celebahq_vqvae.yaml")
+                        dest="config_path", default="config/celebahq.yaml")
     args = parser.parse_args()
     train(args)