The model is about 246 MB on initiation. About 300K pretraining on a customized grayscale set from DFL, then 300-350K Kiril-to-Arnold, Uniform YAM, no flip for either src/dst, batch 6. LRD after 310K. Trained on Geforce 750 Ti.
============================= Model Summary ==============================
== ==
== Model name: LIAE-UD-192-96-32-32-16-PRE-20-9-2022_SAEHDBW ==
== ==
== Current iteration: 347433 ==
== ==
==--------------------------- Model Options ----------------------------==
== ==
== resolution: 192 ==
== face_type: mf ==
== models_opt_on_gpu: True ==
== archi: liae-ud ==
== ae_dims: 96 ==
== e_dims: 32 ==
== d_dims: 32 ==
== d_mask_dims: 16 ==
== masked_training: True ==
== eyes_mouth_prio: True ==
== uniform_yaw: True ==
== blur_out_mask: True ==
== adabelief: True ==
== lr_dropout: cpu ==
== random_warp: True ==
== random_hsv_power: 0.0 ==
== true_face_power: 0.0 ==
== face_style_power: 0.0 ==
== bg_style_power: 0.0 ==
== ct_mode: none ==
== clipgrad: False ==
== pretrain: False ==
== autobackup_hour: 0 ==
== write_preview_history: True ==
== target_iter: 0 ==
== random_src_flip: False ==
== random_dst_flip: False ==
== batch_size: 6 ==
== gan_power: 0.0 ==
== gan_patch_size: 24 ==
== gan_dims: 16 ==
== ==
==----------------------------- Running On -----------------------------==
== ==
== Device index: 0 ==
== Name: 750 Ti ==
== VRAM: 1.45GB ==
== ==
==========================================================================
Note the face in the top-left, LOL:
