Reproducing paper results

[ymohit]

Hello everyone,

Thanks for sharing the code for this awesome research work!

I am having issues in reproducing results mentioned in the paper. For example on running deepGLMM_NormalExample which runs an experiment on the cornwell dataset. I am getting MSE=0.47 while the corresponding number in the paper is 0.05 (which is much better).

I am wondering if I am getting something wrong. I will appreciate any response from authors on this issue. Thanks!

Iteration: 1 - MSE: 1.1446
---------- Start Training Phase ----------
Iteration: 2 - MSE: 1.0386
Iteration: 3 - MSE: 0.95974
Iteration: 4 - MSE: 0.90035
Iteration: 5 - MSE: 0.83946
Iteration: 6 - MSE: 0.75699
Iteration: 7 - MSE: 0.66455
Iteration: 8 - MSE: 0.61805
Iteration: 9 - MSE: 0.61674
Iteration: 10 - MSE: 0.62723
Iteration: 11 - MSE: 0.64137
Iteration: 12 - MSE: 0.65366
Iteration: 13 - MSE: 0.6654
Iteration: 14 - MSE: 0.67801
Iteration: 15 - MSE: 0.69061
Iteration: 16 - MSE: 0.70461
Iteration: 17 - MSE: 0.7184
Iteration: 18 - MSE: 0.73422
Iteration: 19 - MSE: 0.74902
Iteration: 20 - MSE: 0.76529
---------- Training Completed! ----------
Number of iteration:20
MSE best: 0.61674
Training time: 1075.5819s
---------- Prediction ----------
Mean square error on test data: 47.0046