Well, one way would be to read the data documentation. The Scientific Data Hub specifically states: "The Sentinel-2 data offer for the Scientific Data Hub will consist of Level-1C user products", which is in reflectance values.
As such, you may have actually corrupted the data in applying an atmospheric correction that, as part of the processing, corrects to at-sensor reflectance. Please note that the pre-processing tools in the QGIS semi automatic classification plugin are intended for Landsat products and should not have been applied to Sentinel data in the first place. In reading the users manual it is clear that a Dark Object Subtraction (DOS1) atmospheric is applied to the data, but only for Landsat. The only other tools I see under the Pre-processing tab are "Clip Multiple Rasters" and "Split Raster Bands".
Here is the quote from the users manual on pre-processing of Landsat:
The tab Landsat allows for the conversion of Landsat 1, 2, and 3 MSS
and Landsat 4, 5, 7, and 8 images from DN (i.e. Digital Numbers) to
the physical measure of Top Of Atmosphere reflectance (TOA), or the
application of a simple atmospheric correction using the DOS1 method
(Dark Object Subtraction 1), which is an image-based technique (for
more information about the Landsat conversion to TOA and DOS1
correction, see Landsat image conversion to reflectance and DOS1
Is there some other tool that you are using in the toolbox that indicates that an atmospheric correction is applied to your Sentinel-2 data? If so, please edit your post to provide more detail on exactly what you have done.