This is basically in-line with the test standard followed....generally 5 is sufficient as per ASTM. Checking for repeatability from another set of samples is always beneficial.
Roughly speaking, the deviation allowed should not be higher than 10%. However, this will also depend on many factors. First, the length of your fibers (long continuous or short ones). Additionally, when working with fiber reinforced materials, the orientation of these is critical since it induces an anisotropic mechanical behavior of your material. Thus, if you want to validate the repeatability of your results, you must take your samples with same orientation. Otherwise, you are likely having a different values of your modulus depending on your fiber orientation.