@@ -68,7 +68,7 @@ def __init__(
6868 in_channels = input_channels ,
6969 out_channels = conv3d_channels ,
7070 kernel_size = (kernel_size , kernel_size , kernel_size ),
71- padding = (1 ,0 , 0 ),
71+ padding = (1 , 0 , 0 ),
7272 )
7373 )
7474 for i in range (0 , num_layers ):
@@ -77,7 +77,7 @@ def __init__(
7777 in_channels = conv3d_channels ,
7878 out_channels = conv3d_channels ,
7979 kernel_size = (kernel_size , kernel_size , kernel_size ),
80- padding = (1 ,0 , 0 ),
80+ padding = (1 , 0 , 0 ),
8181 )
8282 )
8383
@@ -95,9 +95,7 @@ def __init__(
9595 # Small head model to convert from latent space to PV generation for training
9696 # Input is per-pixel input data, this will be
9797 # reshaped to the same output steps as the latent head
98- self .pv_meta_input = nn .Linear (
99- pv_meta_input_channels , out_features = hidden_dim
100- )
98+ self .pv_meta_input = nn .Linear (pv_meta_input_channels , out_features = hidden_dim )
10199
102100 # Output is forecast steps channels, each channel is a timestep
103101 # For labelling, this should be 1, forecasting the middle
@@ -142,7 +140,5 @@ def forward(self, x: torch.Tensor, pv_meta: torch.Tensor = None, output_latents:
142140 x = torch .cat ([x , pv_meta ], dim = 1 )
143141 # Get pv_meta_output
144142 x = self .pv_meta_output (x )
145- x = F .relu (
146- self .pv_meta_output2 (x )
147- ) # Generation can only be positive or 0, so ReLU
143+ x = F .relu (self .pv_meta_output2 (x )) # Generation can only be positive or 0, so ReLU
148144 return x
0 commit comments