Thanks , have you training tips for SAE model?
i have a GeForce GTX 1050 ti
this is my training
Enable autobackup? n
Write preview history? n
Target iteration: 40000
Batch_size: 8
Feed faces to network sorted by yaw? y
Optimizer mode? 1
Use pixel loss? n
Face style power: 0.0
Background style power: 0.0
Apply random color transfer to src faceset? n
Enable gradient clipping? n
resolution: 128
face type: f
learn mask: y
architecture: df
ae dims : 256
e ch dims : 21
d ch dims: 10
random flip: y
after first 40000 iteration
Enable autobackup? y
Write preview history? n
Target iteration: unlimited
Batch_size: 12
Feed faces to network sorted by yaw? y
Optimizer mode? 1
Use pixel loss? y
Face style power: 0.0
Background style power: 0.0
Apply random color transfer to src faceset? n
Enable gradient clipping? n
resolution: 128
face type: f
learn mask: y
architecture: df
ae dims : 256
e ch dims : 21
d ch dims: 10
random flip: y
is this a good training? change anything?