If you can't find this, what I did with one I repaired is to measure and adjust the bias so the heatsinks after 1hr are 50c on a 25c day, with the lid on.
Not familiar with this amp, but if it has a bipolar output stage, look for emitter resistors. These will all be common to the amp output, with one each to each power transistor. With the power off, measure the resistance with a good DVM with a low ohm scale. Most of these are well under 1 ohm in value. Then with the amp powered up and warmed up, measure the voltage drop across these resistors and compute the current through them from your measured values. Then add up all the emitter resistors (of one polarity) and multiply, and that will give you the total bias current in the output stage. You can then monitor the voltage drop across one of these resistors while you adjust bias to compute the total current.
Sorry I misread your post first time. I think you were asking about what the correct bias current was, not how to measure it.
One thing you can do is determine which power transistor the ML3 uses. Then find out its maximum safe operating area. If you can find a spec sheet, this data is usually published. Then measure the power supply bus voltage for the output stage. Then double it to be safe. Look at the data sheet for the maximum DC current (Ic) the power transistor can have at double the bus voltage without going into secondary breakdown, and this will give you a max top end on the bias current. Of course other parameters such as power dissipation may result in further limitations but you probably should not exceed the maximum DC current at double the bus voltage, just in case you have a failure that pulls the output to one of the rails.
Also, spec sheet says idle power is 180 Watts. You can compute a rough estimate of bias current by determining the bus voltage, multiply by two, and divide into 90 Watts.
So, if you have +/- 90V busses, then that is 180V max, divided into 90 watts give 0.5 amps output stage current. This is a bit of an overestimate because it does not account for the consumption in the rest of the amp, but it may get you close.
I run the same calculation when adjusting the "class A" or high bias current in my JC1s.
One quick thing I thought of after my last post. When I was adjusting bias for the JC1s, I measured the total power dissipation with those digital power meters you can get at the hardware store. You plug your appliance into these and they measure the total power consumed.
You can use this to set the bias so the total power reads Levinson's spec of 180W. This will prevent you from over biasing because it takes into account the consumption in the rest of the amp.