jojonki / pointer-networks Goto Github PK

RuntimeError Traceback (most recent call last)
in
76 if torch.cuda.is_available():
77 model.cuda()
---> 78 train(model, train_X, train_Y, batch_size, n_epochs)
79 print('----Test result---')
80 test(model, test_X, test_Y)

in train(model, X, Y, batch_size, n_epochs)
40 print (type(probs))
41 outputs = probs.view(-1, L) # (bsM, L)
---> 42 y = y.view(-1) # (bsM)
43 loss = F.nll_loss(outputs, y)
44

RuntimeError: view size is not compatible with input tensor's size and stride (at least one dimension spans across two contiguous subspaces). Use .reshape(...) instead.

​