model:
  base_learning_rate: 1.0e-05
  target: ldm.models.diffusion.ddpm_audio.LatentDiffusion_audio
  params:
    linear_start: 0.00085
    linear_end: 0.0120
    num_timesteps_cond: 1
    log_every_t: 200
    timesteps: 1000
    first_stage_key: image
    cond_stage_key: caption
    image_size: 32      # unused
    mel_dim: 10         # 80 // 2^3
    mel_length: 78     # 624 // 2^3
    channels: 4
    cond_stage_trainable: false
    conditioning_key: crossattn
    monitor: val/loss_simple_ema
    scale_by_std: True
    use_ema: False

    scheduler_config: # 10000 warmup steps
      target: ldm.lr_scheduler.LambdaLinearScheduler
      params:
        warm_up_steps: [10000]
        cycle_lengths: [10000000000000]
        f_start: [1.e-6]
        f_max: [1.]
        f_min: [ 1.]

    unet_config:
      target: ldm.modules.diffusionmodules.openaimodel.UNetModel
      params:
        image_size: 32 # ununsed
        in_channels: 4
        out_channels: 4
        model_channels: 320
        attention_resolutions:
        - 1
        - 2
        num_res_blocks: 2
        channel_mult:  # num_down = len(ch_mult)-1
        - 1
        - 2
        num_heads: 8
        use_spatial_transformer: true
        transformer_depth: 1
        context_dim: 1024
        use_checkpoint: true
        legacy: False

    first_stage_config:
      target: ldm.models.autoencoder.AutoencoderKL
      params:
        embed_dim: 4
        monitor: val/rec_loss
        ckpt_path: 
        ddconfig:
          double_z: true
          z_channels: 4
          resolution: 624
          in_channels: 1
          out_ch: 1
          ch: 128
          ch_mult: [ 1, 2, 2, 4 ]  # num_down = len(ch_mult)-1
          num_res_blocks: 2
          attn_resolutions: [78, 156]
          dropout: 0.0
        lossconfig:
          target: torch.nn.Identity

    cond_stage_config:
      target: ldm.modules.encoders.modules.FrozenCLAPEmbedder
      params:
        weights_path: useful_ckpts/CLAP/CLAP_weights_2022.pth

ckpt_path: useful_ckpts/maa1_caps.ckpt