In short, this means they haven't been detected by RepeatModeler. That is because correct detection of SINE's is much harder than of other TEs. EDTA is a different pipeline (but includes RepeatModeler) that has been developed and parameter-tuned to outperform other such pipelines. It is therefore not surprising that it yields slightly different results. You could integrate the SINE sequences into your repeat library, but expect many erroneous classifications. The EDTA paper has more background on this and also elaborates on the areas with room for improvement (including SINE detection).