Accuracy 0 for toy model (Image Classification)

Part 1 (2019)
1

Created a toy model to fit into Learner for image classification, but getting 0 accuracy. Has anyone faced a similar problem before?

model = nn.Sequential(
nn.Linear(24, 36),nn.ReLU(),
nn.Linear(36, 36),nn.ReLU(),
nn.Linear(36, 10),nn.ReLU(),
Lambda(lambda x: x.view(x.size(0),-1))
)

learn = Learner(data, model, metrics=accuracy)

Thanks in advance, Neo

2

Generally I haven’t seen Relu Activation used as activation function in the last layer.
Also can you share some details about the dataset? If it is unbalanced, the network can perform poorly.