Problem in training

[scpaliuhc]

I just increased the value of training_epochs in atn.py and executed the command "python atn.py --train" with a new path for saving the model. But I found that the loss value of each epoch didn't change a lot: Eopch 1 completed. loss = 99.92884826660156 Eopch 2 completed. loss = 98.98079681396484 Eopch 3 completed. loss = 99.99166870117188 Eopch 4 completed. loss = 99.44509887695312 Eopch 5 completed. loss = 98.99491119384766 Eopch 6 completed. loss = 100.26902770996094 Eopch 7 completed. loss = 99.92530822753906 Eopch 8 completed. loss = 99.55029296875 Eopch 9 completed. loss = 99.61663818359375 Eopch 10 completed. loss = 100.05928039550781 Eopch 11 completed. loss = 99.95893859863281 Eopch 12 completed. loss = 97.95149230957031 Eopch 13 completed. loss = 99.51895904541016 Eopch 14 completed. loss = 99.58808898925781 Eopch 15 completed. loss = 101.28242492675781 Eopch 16 completed. loss = 100.20669555664062 Eopch 17 completed. loss = 100.06300354003906 Eopch 18 completed. loss = 101.12290954589844 Eopch 19 completed. loss = 100.303955078125 Eopch 20 completed. loss = 100.51847839355469 Eopch 21 completed. loss = 100.51861572265625 Eopch 22 completed. loss = 100.18065643310547 Eopch 23 completed. loss = 98.83200073242188 Eopch 24 completed. loss = 101.06179809570312 Eopch 25 completed. loss = 99.15919494628906 Eopch 26 completed. loss = 100.65020751953125 Eopch 27 completed. loss = 99.72256469726562 Eopch 28 completed. loss = 99.38571166992188 Eopch 29 completed. loss = 100.00604248046875 Eopch 30 completed. loss = 99.70945739746094

After 30 epochs, I executed "python atn.py" for test, but there is nothing in the images generated by model._autoencoder. All these generated images are pure black.

I only changed the value of epochs and the path that the model will be saved in. How did you train the model you provided in this project？

[kunal266]

same here, can someone help