My training has been hanging around 0.62 for the past day of training. Any suggestions on how to make the training move on past 0.62 or what is happening within the training would be helpful. Thanks! 
3699 0.483304 1.649262 0.625667 00:36 [32/1819]3700 0.462674 1.645753 0.628917 00:37
3701 0.508797 1.641627 0.631833 00:37
3702 0.497441 1.638534 0.628000 00:37
3703 0.509539 1.648153 0.630000 00:37
3704 0.492258 1.648311 0.624417 00:38
3705 0.466166 1.673759 0.622917 00:37
3706 0.484983 1.654282 0.621750 00:37
3707 0.477985 1.650900 0.621083 00:37
3708 0.470231 1.664574 0.622167 00:37
3709 0.485027 1.641483 0.626583 00:37
3710 0.519759 1.643189 0.626083 00:38
3711 0.486719 1.633917 0.624333 00:37
3712 0.488696 1.644063 0.628417 00:37
3713 0.466371 1.656702 0.628750 00:37
3714 0.493300 1.662844 0.625417 00:38
3715 0.492837 1.655735 0.623583 00:38
3716 0.490509 1.671735 0.624167 00:37
3717 0.482492 1.660005 0.623500 00:38
3718 0.519999 1.643638 0.625083 00:37
3719 0.493484 1.638869 0.628250 00:38
3720 0.473004 1.642064 0.624750 00:37
3721 0.498152 1.647380 0.625917 00:37
3722 0.491993 1.645479 0.626083 00:37
3723 0.485566 1.650342 0.624833 00:38
3724 0.519620 1.640946 0.627333 00:38
3725 0.464361 1.613261 0.627083 00:38
3726 0.490036 1.660096 0.625417 00:38
3727 0.491244 1.645415 0.624500 00:37
3728 0.461206 1.664434 0.628000 00:37
3729 0.512688 1.637055 0.627583 00:37
path = Path(untar_data(URLs.CIFAR_100))
learning_rate = 0.0005133
dropout = 0.4068
bs = 32
epochs=40000
model = models.resnet50
tfms = get_transforms(do_flip=True, flip_vert=True, max_rotate=25, max_lighting=0.1, max_zoom=1.05, max_warp=0.9)
data = (ImageList.from_folder(path)
.split_by_rand_pct()
.label_from_folder()
.transform(get_transforms())
.databunch()
.normalize(imagenet_stats))
learn = cnn_learner(data,
model,
pretrained=True,
ps=dropout,
metrics=accuracy)
learn.fit_one_cycle(epochs, max_lr=learning_rate)
learn.save(‘inity1’)
learn.export(‘init1y.pkl’)