Deep Learning Hands-on

Following

Session 2 & 3

from torch.nn.parallel import data_parallel

This line seems to be automatically added to my code after the first run.

Session 4

# Final size: 1000, 20, 1
# Default LSTM layer input: sequence length, batch size, input size

The scale doesn't match seemingly.

The test set is 100 same sine waves. But I found out that the 100 results seem to be different when I print them.