When i was training my model, I found the loss doesn’t decrease, My model input is 3-D image, and the loss function is binary cross entropy. The optimizer is SGD with momentum 0.9, initial learning rate 0.01, batch size 256.
the loss in the 30 epochs is
0.749619674
0.693235771, 0.69320591, 0.693198329, 0.693196385, 0.693205666, 0.693200624
0.693216185, 0.69323178, 0.693249031, 0.69317602, 0.693203645, 0.69319184
0.693239689, 0.693188139, 0.693181048, 0.693167897, 0.693213663, 0.693183094
0.693211535, 0.693212263, 0.69321184, 0.693209582, 0.693231421, 0.693211854
0.693185517, 0.693207201, 0.693201362, 0.693189077, 0.693243751
Anyone can give me a suggestion? Thanks!