Result of resnet18 on imagenet is low until epoch60

[Jilliansea]

Hi, I use FPGM-only to train resnet18 on imagenet with 'python pruning_imagenet.py /nfs/volume-92-4/data/ILSVRC2012 -a resnet18 --save_dir ./snapshots/resnet18-rate-0.3 --rate_norm 1 --rate_dist 0.3 --layer_begin 0 --layer_end 57 --layer_inter 3 --workers 12 --resume ./snapshots/resnet18-rate-0.3'，
And until now it has trained 60 epochs， the best result is top1=59.634, top5=82.898 on the test set, which is too low than the official result, is it normal?