I ran this code to replicate the results reported in the research paper.
on MR I got 75% instead of 77% reported in paper
on R8 I got 95% instead of 98%
on R52 I got 89.5% instead of 95%.
Do I need to perform any additional changes in the code to get the same results.