Hello,
How can you explain the following phenomenon, loss increases drastically and performance gets better. l’m supposed to get bad performance, isn’t it ?
epoch= 1, loss(train)= 26.956, accuracy(train)= 54.545, time= 1.786, lr= 0.00010
accuracy(test) = 0.000 %, time= 0.558
epoch= 2, loss(train)= 42.272, accuracy(train)= 81.818, time= 1.044, lr= 0.00010
accuracy(test) = 50.000 %, time= 0.586
epoch= 3, loss(train)= 7.075, accuracy(train)= 97.727, time= 1.032, lr= 0.00009
accuracy(test) = 50.000 %, time= 0.535
epoch= 4, loss(train)= 116.282, accuracy(train)= 93.182, time= 1.045, lr= 0.00009
accuracy(test) = 50.000 %, time= 0.537
epoch= 5, loss(train)= 1721.272, accuracy(train)= 86.364, time= 0.946, lr= 0.00008
accuracy(test) = 50.000 %, time= 0.497
epoch= 6, loss(train)= 2008.031, accuracy(train)= 86.364, time= 1.002, lr= 0.00008
accuracy(test) = 75.000 %, time= 0.545