The RSME of the autoencoder is not reducing after training

When I train for the first autoencoder using the code below, the Mean Squared Error is not reducing. However when I did it for the second autoencoder (with different epoch), it performed well with a reducing Mean Sqeared Error.

hiddenSize1 = 100
autoenc1 = trainAutoencoder(train1,hiddenSize1, ...
    'MaxEpochs',400, ...
    'L2WeightRegularization',0.004, ...
    'SparsityRegularization',4, ...
    'SparsityProportion',0.15, ...
    'ScaleData', false);

[yanqi liu]

may change L2WeightRegularization to check

[Mahesh Taparia]

What other parameters you have changed for second autoencoder a part from number of epoch?

[Ng Yong Jie]

I did not change anything for the seccond autoencoder.

[Mahesh Taparia]

Can you share the autoencoder objects and plot for both the cases?