This rate would be variable according to measurement method used. For example, in older people, Latham et al observed an increased of + 20% to +200% for 1 RM, and +10% to +30% for isokinetic measure.
For triceps surae, Weiss et al observed an increased of 13% of isotonic plantar flexion strength after a 24 sessions training program.
You probably could found more indication in these studies, regarding figures and tables results.
This is my second-thought answer. If you include in the consideration the cases when the muscles are forcibly stretched, like in takeoffs in high and long jumping, you will get extremely large values. For instance elite long jumpers exert a GRF that exceeds 10-12 body weights (BW) during the first peak and 4-5 BW during the second peak. The time of contact can be as small as 120 ms. Assuming for estimation purposes that the max force is exerted at 50% of the ground contact time and dividing 10 BW by 50 ms you will obtain extremely high values of the rate of force development (even without transforming the GRF values into the gastrocnemius and soleus forces).
Vladimir Zatsiorsky is completely right in stating that in eccentric contractions dF/dt can be very fast. Mind that a muscle is composed of a contractile and an elastic element in series. For fast eccentric actions dF/dt = dF/dx . dx/dt = k.dx/dt in which k is the elastic stiffness, thus dF/dt mainly depends on the lengthening speed dx/dt, not on the muscle contractile properties.
For comparing muscle contractile properties, you may better use isometric actions. For this case you can find some formulas in my paper 'Muscle mechanics and neuromuscular control, J. Biomechanics 36:1031-1038 (2003) in Appendix 1.