This won't be an "efficiency curve". For a given engine it will be a complex multidimensional map that is a function of the mechanical design of the engine and its calibration and, perhaps more significantly than anything else, its operating conditions.
The one that's most of interest is the mechanical power out relative to the chemical energy in. That is called the "brake specific fuel consumption" (BSFC). Look up BSFC charts of a few engines and you'll see what I mean. They take the form of a chart with the main variables RPM on one axis and torque or BMEP (brake mean effective pressure) on the other axis and you look up what the fuel consumption is on that chart ... except that this one chart will be particular to specified barometric pressure, temperature, coolant temperature, etc., and if you change those things (significantly), it changes the chart. This is NOT simple. For a given engine, the RPM and the BMEP are the two main influencing factors but the others are not zero.
As for what portion goes out the tailpipe and what portion goes out the radiator ... it won't be completely out of the ballpark to estimate that whatever energy input doesn't come out as shaft power, is split more-or-less evenly between cooling system and exhaust pipe, with the chemical-energy lost to the exhaust negligible on a modern engine if the catalyst is taken into account as being part of the complete engine. But ... so what? Neither are useful to you aside from heating the interior in winter, and perhaps being necessary to size the cooling system components. Design features such as integral exhaust manifolds cast into the cylinder head tip the balance towards the heat going to the radiator (and making the engine warm up faster) because they're surrounded by cooling jackets inside the cylinder head.