reproduction of Micronet-M0 4GPU

[HydrogenSulfate]

Thanks for your great work, but I have trained the train_micronet_m0_4gpu.sh script, but got best Acc@1 45.86 at epoch 596, lower than 46.6 reported in your repo with pytorch 1.7.1 CUDA 10.2 cudnn 7.6.05.

And I did not make any modifications to the training program, is there something wrong with my training program?

[liyunsheng13]

The smallest model does have some stability issues, but it should have less than 0.5% variance. I'm curious whether you can reproduce the result with the released model.

[HydrogenSulfate]

The smallest model does have some stability issues, but it should have less than 0.5% variance. I'm curious whether you can reproduce the result with the released model.

the accuracy of the provided model is ok(46.69).

[HydrogenSulfate]

The smallest model does have some stability issues, but it should have less than 0.5% variance. I'm curious whether you can reproduce the result with the released model.

one more question, why the learning rate is 10 times larger than descripted in paper?

[flytocc]

How about the stability of the largest model (m3)? @liyunsheng13

[liyunsheng13]

The largest model should be stable. In the experiments I did, the larger the model is, the stabler the performance will be.