Calculating accuracy on RNN-LSTM model

[hobbyist]

Hello,

How can I calculate the accuracy in a RNN-LSTM neural network?
Also when testing my model with either epoch = 1 , or epoch = 40 the result of the loss (0,01...) is approximately the same and I want to be more specific and calculate the accuracy in case there are errors with "loss" calculation.

Thanks

update: I changed this line:

model.compile(loss='mean_squared_error', optimizer='adam')

with:

model.compile(loss='mean_squared_error', optimizer='adam',metrics=['accuracy'])

and I get accuracy such as this:

...accuracy: 1.9380e-04 

or this:

...accuracy: 2.9070e-04

Update2:
Wtih this command:

model.compile(loss='mean_squared_error', optimizer='adam', metrics=[tensorflow.keras.metrics.Accuracy()])

I get this output:

...accuracy: 0.0000e+00

which does not seem good... any idea?