Cannot reproduce the results on the PanNuke dataset

[mqbin]

Hello! I am trying to reproduce the results in the paper PanNuke Dataset Extension, Insights and Baselines. I used the following dataset split :Training: Fold 1; Validation: Fold 2; Testing: Fold 3.
I used the PyTorch version of the code for training and obtained the following results:

mean,0.4213653276623722,0.6425824540937765(PQ, PQ bin)

The obtained results are lower than the average results mentioned in the paper.I used the fast model mode and commented out the affine transformation.