I think you meant "big" rather than "hot"? It's decay rate is low enough that it doesn't even melt itself when not cooled (and it melts at room temperature). Adding precision here, a 'thin layer' would not melt itself, you could probably keep a 10+kg chunk liquid. But the low efficiency means you need a lot of surface area between the Cs137 and the silicon lattice, so it one could calculate the number of sq meters of "battery" surface you would need to power a device and then see if you could topologically fold that into a relatively small area. Cs137 also can produce a small amount of gamma radiation which you'd want to shield against so the whole thing might end up in a lead box. (not great for making it light :-))

It would be a fun engineering challenge and a good use of nuclear waste but difficult to sell politically because the association with "nuclear." And yes, it isn't actually practical at this level. Looking at where folks are looking at this seriously it seems to be more of a spacecraft thing (long duration probes far from the Sun kind of thing)