Hidden_size in torch.nn.LSTM

Hi guys, I’m starting to learn about LSTM recently and I don’t quite understand about hidden_size in torch.nn.LSTM. Can anyone help me to understand this?