Hello,
I want to use my NN as inference. for some NN’s this works fine but for other i get immideate error when pressing the play bottun.
Command :
Error :
Someone know what could be the reason for this error ?
I will note that for some NN’s it works but for others not.
Thanks !
Hi,
For next time, please copy paste the stack rather than use a screenshot. Thank you.
One of the errors reads :
size mismatch for critic.network_body.linear_encoder.seq_layers.2.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([256]).
This means that the model that you are trying to load the checkpoint from has a different architecture than the network you are trying to use. When using --initialize-from, you need to make sure the network created for the training session has the same architecture as the network you are initializing from. Here, I suspect the network layer size went from 128 to 256, preventing initialization.
