size of sequence input layer of LSTM

I am working on bi LSTM.I have dataset comprises of 1720 samples with 6 features.When i set the size of sequence input layer (size of cell array) to 1, it gives me good accuracy but as i increases the size of input layer to 5 and 10, the accuracy decreases and training time also reduces. Kindly explain me why the size of input layer in LSTM has an affect on Accuracy?

Answer 1

[Dario Walter]

Because the inputSize is equal to the number of used features (6 in your case). See documentation https://de.mathworks.com/help/deeplearning/ref/nnet.cnn.layer.sequenceinputlayer.html