def fcnn_train(train_input, train_output):
clf = FullyConnectedNN()
# print(clf)
fcnn_learner = Learner(data=data,
model=clf,
loss_func=nn.CrossEntropyLoss(),
metrics=accuracy)
fcnn_learner.fit_one_cycle(5, 1e-2)
def fcnn_train(train_input, train_output, test_input, test_output):
tensor_train_input = torch.from_numpy(train_input)
tensor_train_output = torch.from_numpy(train_output[:, 1].astype(int))
tensor_test_input = torch.from_numpy(test_input)
tensor_test_output = torch.from_numpy(test_output[:, 1].astype(int))
train_ds = ArrayDataset(tensor_train_input, tensor_train_output)
test_ds = ArrayDataset(tensor_test_input, tensor_test_output)
bs = 10
databunch = DataBunch.create(train_ds, test_ds, bs=bs)
clf = FullyConnectedNN()
# print(clf)
fcnn_learner = Learner(data=databunch,
model=clf,
loss_func=nn.CrossEntropyLoss(),
metrics=accuracy)
fcnn_learner.fit_one_cycle(50, 1e-2)
# breakpoint()
return fcnn_learner