ShreeDevi,
Where did this training text come from? It includes two different Georgian scripts (mkhedruli and asomtavruli). Only mkhedruli is in common usage today, so it seems to me that it would be best to remove the asomtavruli to increase accuracy on modern texts. If complete historical accuracy is desired, then the third Georgian script (nuskhuri) should probably be included as well.
In addition, you might try contacting Levan Gelashvili (CCed), who has created a tesseract-based OCR program for Georgian; I haven't had very good results with SunnyPage, but he may have improved it since the last time I tried it.