config_coco.yaml

model:
  base_learning_rate: 1.0e-06
  target: ldm.models.diffusion.ddpm.LatentDiffusion
  params:
    linear_start: 0.0015
    linear_end: 0.0195
    num_timesteps_cond: 1
    log_every_t: 200
    timesteps: 1000
    first_stage_key: image
    cond_stage_key: class_label
    image_size: 32
    channels: 4
    cond_stage_trainable: false
    conditioning_key: crossattn
    cond_stage_forward: encode_graph_local_global
    monitor: val/loss_simple_ema
    unet_config:
      target: ldm.modules.diffusionmodules.openaimodel.UNetModel
      params:
        image_size: 32
        in_channels: 4
        out_channels: 4
        model_channels: 256
        attention_resolutions:
        - 4
        - 2
        - 1
        num_res_blocks: 2
        channel_mult:
        - 1
        - 2
        - 4
        num_head_channels: 32
        use_spatial_transformer: true
        transformer_depth: 1
        context_local_dim: 1536
        context_dim: 512
    first_stage_config:
      target: ldm.models.autoencoder.VQModelInterface
      params:
        embed_dim: 4
        n_embed: 16384
        ckpt_path: pretrained/vq-f8-model.ckpt
        ddconfig:
          double_z: false
          z_channels: 4
          resolution: 256
          in_channels: 3
          out_ch: 3
          ch: 128
          ch_mult:
          - 1
          - 2
          - 2
          - 4
          num_res_blocks: 2
          attn_resolutions:
          - 32
          dropout: 0.0
        lossconfig:
          target: torch.nn.Identity
    cond_stage_config:
      target: ldm.modules.cgip.cgip.CGIPModel
      params:
        num_objs: 184
        num_preds: 7
        layers: 5
        width: 512
        embed_dim: 512
        ckpt_path: pretrained/sip_coco.pt
data:
  target: scripts.datamodule.DataModuleFromConfig
  params:
    batch_size: 16
    num_workers: 4
    wrap: false
    train:
      target: ldm.data.coco.COCOTrain
      params:
        image_dir: ./datasets/coco/images/train2017
        instances_json: ./datasets/coco/annotations/instances_train2017.json
        stuff_json: ./datasets/coco/annotations/stuff_train2017.json
        stuff_only: True
        image_size: 256
    validation:
      target: ldm.data.coco.COCOValidation
      params:
        image_dir: ./datasets/coco/images/val2017
        instances_json: ./datasets/coco/annotations/instances_val2017.json
        stuff_json: ./datasets/coco/annotations/stuff_val2017.json
        stuff_only: True
        image_size: 256

lightning:
  callbacks:
    image_logger:
      target: trainer.ImageLogger
      params:
        batch_frequency: 10000
        max_images: 16
        increase_log_steps: False
  trainer:
    benchmark: True