Jul 4, 2022
I ran your example using cpu and after epoch 10 the result was
Epoch 10
-------------------------------
[10, 500/3125] loss: 0.435
[10, 1000/3125] loss: 0.435
[10, 1500/3125] loss: 0.437
[10, 2000/3125] loss: 0.430
[10, 2500/3125] loss: 0.424
[10, 3000/3125] loss: 0.430
Test Error:
Accuracy: 39.4%, Avg loss: 1.683843