Hi guys, I am working on a OCR model. My vaidation accuracy is near 80 percent but still test predictions are very wrong. Not able to predict even one. Any suggestions?
