georgefen/Face-Landmark-ControlNet
116
1model:2 target: cldm.cldm.ControlLDM3 params:4 linear_start: 0.000855 linear_end: 0.01206 num_timesteps_cond: 17 log_every_t: 2008 timesteps: 10009 first_stage_key: "jpg"10 cond_stage_key: "txt"11 control_key: "hint"12 image_size: 6413 channels: 414 cond_stage_trainable: false15 conditioning_key: crossattn16 monitor: val/loss_simple_ema17 scale_factor: 0.1821518 use_ema: False19 only_mid_control: False20 21 control_stage_config:22 target: cldm.cldm.ControlNet23 params:24 image_size: 32 # unused25 in_channels: 426 hint_channels: 327 model_channels: 32028 attention_resolutions: [ 4, 2, 1 ]29 num_res_blocks: 230 channel_mult: [ 1, 2, 4, 4 ]31 num_heads: 832 use_spatial_transformer: True33 transformer_depth: 134 context_dim: 76835 use_checkpoint: True36 legacy: False37 38 unet_config:39 target: cldm.cldm.ControlledUnetModel40 params:41 image_size: 32 # unused42 in_channels: 443 out_channels: 444 model_channels: 32045 attention_resolutions: [ 4, 2, 1 ]46 num_res_blocks: 247 channel_mult: [ 1, 2, 4, 4 ]48 num_heads: 849 use_spatial_transformer: True50 transformer_depth: 151 context_dim: 76852 use_checkpoint: True53 legacy: False54 55 first_stage_config:56 target: ldm.models.autoencoder.AutoencoderKL57 params:58 embed_dim: 459 monitor: val/rec_loss60 ddconfig:61 double_z: true62 z_channels: 463 resolution: 25664 in_channels: 365 out_ch: 366 ch: 12867 ch_mult:68 - 169 - 270 - 471 - 472 num_res_blocks: 273 attn_resolutions: []74 dropout: 0.075 lossconfig:76 target: torch.nn.Identity77 78 cond_stage_config:79 target: ldm.modules.encoders.modules.FrozenCLIPEmbedder80