The amount of lift is hard to estimate, since it's a function of airspeed and angle of attack, modified by the fact that the wing is functioning in ground effect. A back of envelope sum says that it'll reduce by something around the weight multiplied by the square root of (speed/stalling speed), but that value could be as much as 50% out and I wouldn't hang my hat on it.
That estimate would be based on the assumption that the AOA with the aircraft rolling on all three wheels is equal to the critical (stalling) AOA. That would be an extreme case and would make the aircraft very difficult to land on the mains. As G-ALAN suggests, because the aircraft has to rotate at Vr, the AOA while rolling is less than the critical AOA.
Some aircraft (and I'm ignoring "conventional" undercarriage here!) sit at a higher AOA than others. The Twin Comanche was designed to allow the passengers to enplane without the use of a step, hence it has a very nose-up resting attitude. In my experience, this makes it somewhat trying in crosswinds, as you have very little weight left on the wheels late in the take-off roll.
Others -- and it tends to be the high wing ones, though I'm not sure I know why -- have a very low lift coefficient while rolling along the runway, and have to be hauled fairly vigorously into the air.