mnist数据集mlp-noniid的运行结果

[Chrisable]

请问一下大佬，为什么mlp-noniid-mnist第一次测试集运行结果是75%，第二次运行就78%甚至83%+？变化这么大的原因是什么？

第一次结果：

Round   0, Average loss 0.133
Round   1, Average loss 0.097
Round   2, Average loss 0.084
Round   3, Average loss 0.063
Round   4, Average loss 0.075
Round   5, Average loss 0.057
Round   6, Average loss 0.041
Round   7, Average loss 0.049
Round   8, Average loss 0.076
Round   9, Average loss 0.056
Training accuracy: 74.83
Testing accuracy: 75.21

第二次结果：

Round   0, Average loss 0.128
Round   1, Average loss 0.068
Round   2, Average loss 0.099
Round   3, Average loss 0.060
Round   4, Average loss 0.057
Round   5, Average loss 0.070
Round   6, Average loss 0.069
Round   7, Average loss 0.057
Round   8, Average loss 0.066
Round   9, Average loss 0.049
Training accuracy: 78.18
Testing accuracy: 78.39

[Pnme79]

（python main_fed.py --dataset mnist --iid --num_channels 1 --model mlp --epochs 10 --gpu -1）我的就更离谱 FedAVG-MLP Acc. of IID
运行结果：
Round 0, Average loss 2.269
Round 1, Average loss 2.254
Round 2, Average loss 2.261
Round 3, Average loss 2.253
Round 4, Average loss 2.251
Round 5, Average loss 2.246
Round 6, Average loss 2.242
Round 7, Average loss 2.247
Round 8, Average loss 2.239
Round 9, Average loss 2.239
Training accuracy: 11.98
Testing accuracy: 10.56
是什么原因啊？