torchsummaryM
- Just for myself
- Visualize kernel size, input & output shapes and # parameters
- Can handle, RNN. Recursice Net, even multi input model
Ref
Usage
- input_size must be a tuple without batch size
- batch size for calculating memory
- device type one of 'cuda' or 'cpu'
- dtypes for RNN ==> torch.long
summary(model, tensor, *args)