Lstm model prediction converging to a value

Hi, I’ve trained an lstm model on a time seres prediciotn task and when I use it to predict the whole time series it almost match the original, so it’s pretty accurate. But when I use it to predict one step at a time, appending the last prediction to the data and using the last “window size” item count the model predict values converging to a certain value, depending on the first input. I’ve also tried checking if the prediction on one single window size elements count and the whole dataset at the specified index is the same or similar, but it’s not. Any ideas? Thanks in advance