I took a look at the schematic, and as JJ and TLS indicated the design intentionally provides different sensitivities on the two sets of inputs, to provide flexibility for differing applications.
Zd, looking at the schematic I see that in this particular case there is no difference in the active circuitry that is used for the two sets of inputs. The only difference is in the configuration of some resistors and pots (potentiometers, i.e., variable resistors) that are located right at the inputs of the amp.
As to which set of inputs is best to use, if in fact it makes any difference, it's probably not predictable and would have to be determined experimentally. It depends on various design characteristics of the preamp, including what part of the range of its volume control it performs best at; on the output level(s) of the source component(s); on whether or not any ground loop effects occur between the preamp and power amp, that might cause or contribute to low frequency hum or high frequency buzz (the severity of which might be reduced by increasing the signal level between the components); and on the condition of the 50+ year old pots in the power amp. There are pots on both sets of inputs, btw, a balance control being provided on the low (2 volt) sensitivity inputs, and independent gain controls being provided for each channel on the high sensitivity inputs.