I follow the example directly from the website by copy-pasting the code but it seems like the network is resetting the weight after each epoch
https://mxnet.incubator.apache.org/tutorials/python/mnist.html
Below is output produced by the example code from the website to run mnist
INFO:root:Epoch[0] Batch [100] Speed: 59091.52 samples/sec accuracy=0.100000
INFO:root:Epoch[0] Batch [200] Speed: 103752.63 samples/sec accuracy=0.098300
INFO:root:Epoch[0] Batch [300] Speed: 149265.08 samples/sec accuracy=0.097700
INFO:root:Epoch[0] Batch [400] Speed: 148894.17 samples/sec accuracy=0.094500
INFO:root:Epoch[0] Batch [500] Speed: 152394.90 samples/sec accuracy=0.107300
INFO:root:Epoch[0] Train-accuracy=0.094444
INFO:root:Epoch[0] Time cost=0.546
INFO:root:Epoch[0] Validation-accuracy=0.098000
INFO:root:Epoch[1] Batch [100] Speed: 145776.78 samples/sec accuracy=0.100000
INFO:root:Epoch[1] Batch [200] Speed: 130015.22 samples/sec accuracy=0.098300
INFO:root:Epoch[1] Batch [300] Speed: 147538.70 samples/sec accuracy=0.097700
INFO:root:Epoch[1] Batch [400] Speed: 144358.88 samples/sec accuracy=0.094500
INFO:root:Epoch[1] Batch [500] Speed: 134365.21 samples/sec accuracy=0.107300
INFO:root:Epoch[1] Train-accuracy=0.094444
INFO:root:Epoch[1] Time cost=0.435
INFO:root:Epoch[1] Validation-accuracy=0.098000
INFO:root:Epoch[2] Batch [100] Speed: 135762.60 samples/sec accuracy=0.100000
INFO:root:Epoch[2] Batch [200] Speed: 141163.42 samples/sec accuracy=0.098300
INFO:root:Epoch[2] Batch [300] Speed: 147305.53 samples/sec accuracy=0.097700
INFO:root:Epoch[2] Batch [400] Speed: 141288.96 samples/sec accuracy=0.094500
INFO:root:Epoch[2] Batch [500] Speed: 138121.37 samples/sec accuracy=0.107300
INFO:root:Epoch[2] Train-accuracy=0.094444
INFO:root:Epoch[2] Time cost=0.433
INFO:root:Epoch[2] Validation-accuracy=0.098000
INFO:root:Epoch[3] Batch [100] Speed: 136330.52 samples/sec accuracy=0.100000
INFO:root:Epoch[3] Batch [200] Speed: 142017.08 samples/sec accuracy=0.098300
INFO:root:Epoch[3] Batch [300] Speed: 142116.21 samples/sec accuracy=0.097700
INFO:root:Epoch[3] Batch [400] Speed: 140498.58 samples/sec accuracy=0.094500
INFO:root:Epoch[3] Batch [500] Speed: 137093.40 samples/sec accuracy=0.107300
INFO:root:Epoch[3] Train-accuracy=0.094444
INFO:root:Epoch[3] Time cost=0.430
INFO:root:Epoch[3] Validation-accuracy=0.098000
INFO:root:Epoch[4] Batch [100] Speed: 135096.61 samples/sec accuracy=0.100000
INFO:root:Epoch[4] Batch [200] Speed: 142692.04 samples/sec accuracy=0.098300
INFO:root:Epoch[4] Batch [300] Speed: 132139.02 samples/sec accuracy=0.097700
INFO:root:Epoch[4] Batch [400] Speed: 134674.98 samples/sec accuracy=0.094500
INFO:root:Epoch[4] Batch [500] Speed: 136128.32 samples/sec accuracy=0.107300
INFO:root:Epoch[4] Train-accuracy=0.094444
INFO:root:Epoch[4] Time cost=0.440
INFO:root:Epoch[4] Validation-accuracy=0.098000
INFO:root:Epoch[5] Batch [100] Speed: 149180.14 samples/sec accuracy=0.100000
INFO:root:Epoch[5] Batch [200] Speed: 143346.98 samples/sec accuracy=0.098300
INFO:root:Epoch[5] Batch [300] Speed: 136939.43 samples/sec accuracy=0.097700
INFO:root:Epoch[5] Batch [400] Speed: 136022.81 samples/sec accuracy=0.094500
INFO:root:Epoch[5] Batch [500] Speed: 144179.25 samples/sec accuracy=0.107300
INFO:root:Epoch[5] Train-accuracy=0.094444