beam search groups for summarization

beam search groups for summarization just seem to take.. longer?

  "The `device` argument is deprecated and will be removed in v5 of Transformers.", FutureWarning
	In this chapter, the focus is on removing trends and seasonality from time series. Decomposition approaches are not appropriate for all time series because of the non-stationarity of some time series; instead, differencing approaches can be used to remove trend and seasonality. The first step in removing trend is to take differences at lagged xt minus 1. This removes the trend but does not change the remainder term, so it is called a "wide noise series." On the other hand, if there is an exact linear trend function, then the trend will be removed but the mean will not be zero, which is called the drift term.
Score:	-2.4814
	In this chapter, we're going to look at a few different ways to remove the trend and seasonal effects from time series. First, we can use differencing to get rid of the seasonally-varying part of the series. This doesn't always work, though, because it might not remove enough of the trend. The next approach is to smooth out the time series with a moving mean or a filtering function. We'll see how that works in a moment, but for now, just know that you need to include at least two complete periods of observations to get an accurate trend estimate.
Score:	-2.5502