config.yaml

model:
  target: simgen.models.cascade_controlnet.UniControlNet
  params:
    linear_start: 0.00085
    linear_end: 0.0120
    num_timesteps_cond: 1
    log_every_t: 200
    timesteps: 1000
    first_stage_key: "jpg"
    cond_stage_key: "txt"
    image_size: 64
    channels: 4
    cond_stage_trainable: false
    conditioning_key: crossattn
    monitor: val/loss_simple_ema
    scale_factor: 0.18215
    use_ema: False
    mode: local
    parameterization: "v"

    local_control_config:
      target: simgen.models.local_adapter.LocalAdapter
      params:
        in_channels: 4
        model_channels: 320
        local_channels: 6 # 21, then 6 for 2 condition, now 15 for 5
        inject_channels: [192, 256, 384, 512]
        inject_layers: [1, 4, 7, 10]
        num_res_blocks: 2
        attention_resolutions: [4, 2, 1]
        channel_mult: [1, 2, 4, 4]
        use_checkpoint: True
        # num_heads: 8
        num_head_channels: 64 # need to fix for flash-attn
        use_spatial_transformer: True
        use_linear_in_transformer: True
        transformer_depth: 1
        context_dim: 1024 # 768
        legacy: False

    unet_config:
      target: simgen.models.local_adapter.LocalControlUNetModel
      params:
        image_size: 32
        in_channels: 4
        model_channels: 320
        out_channels: 4
        num_res_blocks: 2
        attention_resolutions: [4, 2, 1]
        channel_mult: [1, 2, 4, 4]
        use_checkpoint: True
        # num_heads: 8
        num_head_channels: 64 # need to fix for flash-attn
        use_spatial_transformer: True
        use_linear_in_transformer: True
        transformer_depth: 1
        context_dim: 1024 # 768
        legacy: False

    first_stage_config:
      target: simgen.ldm.models.autoencoder.AutoencoderKL
      params:
        embed_dim: 4
        monitor: val/rec_loss
        ddconfig:
          double_z: true
          z_channels: 4
          resolution: 256
          in_channels: 3
          out_ch: 3
          ch: 128
          ch_mult:
          - 1
          - 2
          - 4
          - 4
          num_res_blocks: 2
          attn_resolutions: []
          dropout: 0.0
        lossconfig:
          target: torch.nn.Identity

    cond_stage_config:
      target: simgen.ldm.modules.encoders.modules.FrozenOpenCLIPEmbedder
      params:
        freeze: True
        layer: "penultimate"
      # target: ldm.modules.encoders.modules.FrozenCLIPEmbedder

first_cond_config:
  target: simgen.models.t2i_model.T2IModel
  params:
    linear_start: 0.00085
    linear_end: 0.0120
    num_timesteps_cond: 1
    log_every_t: 200
    timesteps: 1000
    first_stage_key: "jpg"
    cond_stage_key: "txt"
    image_size: 64
    channels: 4
    cond_stage_trainable: false
    conditioning_key: crossattn
    monitor: val/loss_simple_ema
    scale_factor: 0.18215
    use_ema: False
    parameterization: "v"

    unet_config:
      target: simgen.ldm.modules.diffusionmodules.openaimodel.UNetModel
      params:
        image_size: 32
        in_channels: 4
        model_channels: 320
        out_channels: 4
        num_res_blocks: 2
        attention_resolutions: [4, 2, 1]
        channel_mult: [1, 2, 4, 4]
        use_checkpoint: True
        # num_heads: 8
        num_head_channels: 64 # need to fix for flash-attn
        use_spatial_transformer: True
        use_linear_in_transformer: True
        transformer_depth: 1
        context_dim: 1024 # 768
        legacy: False

    first_stage_config:
      target: simgen.ldm.models.autoencoder.AutoencoderKL
      params:
        embed_dim: 4
        monitor: val/rec_loss
        ddconfig:
          double_z: true
          z_channels: 4
          resolution: 256
          in_channels: 3
          out_ch: 3
          ch: 128
          ch_mult:
          - 1
          - 2
          - 4
          - 4
          num_res_blocks: 2
          attn_resolutions: []
          dropout: 0.0
        lossconfig:
          target: torch.nn.Identity

    cond_stage_config:
      target: simgen.ldm.modules.encoders.modules.FrozenOpenCLIPEmbedder
      params:
        freeze: True
        layer: "penultimate"
      # target: ldm.modules.encoders.modules.FrozenCLIPEmbedder