DF-udt_384_wf_100万迭代次数_高质量模型
模型参数如下，质量也是不错的，可自行下载================= Model Summary =================
== ==
== Model name: p384dfudt_SAEHD ==
== ==
== Current iteration: 1000000 ==
== ==
==--------------- Model Options ---------------==
== ==
== archi: df-udt ==
== resolution: 384 ==
== ae_dims: 384 ==
== e_dims: 84 ==
== d_dims: 84 ==
== d_mask_dims: 28 ==
== face_type: wf ==
== session_name: ==
== autobackup_hour: 0 ==
== maximum_n_backups: 0 ==
== write_preview_history: False ==
== pretrain: True ==
== use_fp16: False ==
== target_iter: 1000000 ==
== batch_size: 8 ==
== models_opt_on_gpu: True ==
== masked_training: True ==
== eyes_prio: True ==
== mouth_prio: True ==
== uniform_yaw: True ==
== blur_out_mask: False ==
== adabelief: True ==
== lr_dropout: n ==
== loss_function: SSIM ==
== lr: 5e-05 ==
== clipgrad: False ==
== retraining_samples: False ==
== background_power: 0.0 ==
== true_face_power: 0.0 ==
== face_style_power: 0.0 ==
== bg_style_power: 0.0 ==
== eyes_mouth_prio: True ==
== cpu_cap: 8 ==
== preview_samples: 8 ==
== force_full_preview: True ==
== gan_version: 2 ==
== gan_power: 0.0 ==
== gan_patch_size: 64 ==
== gan_dims: 16 ==
== gan_smoothing: 0.1 ==
== gan_noise: 0.0 ==
== random_warp: False ==
== random_src_flip: False ==
== random_dst_flip: False ==
== random_downsample: False ==
== random_noise: False ==
== random_blur: False ==
== random_jpeg: False ==
== random_hsv_power: 0.0 ==
== ct_mode: none ==
== random_color: False ==
== src_pak_name: faceset ==
== dst_pak_name: faceset ==
== random_shadow: none ==
== ==
==---------------- Running On -----------------==
== ==
== Device index: 0 ==
== Name: Tesla P100-PCIE-16GB ==
== VRAM: 14.61GB ==
== ==
=================================================
***本内容需购买可见***
:lol 效果挺好的 大家可以试一下啦 OOM报错， 这个板块所有模型都没有预览图 感谢分享 没有预览图！！！！！！ 感谢分享 不懂怎么玩，灵石好不容易过了300，今天不知为啥没了很多
页:
[1]