{
  "model_config": {
    "RGB_normalizer": 0.18215,
    "arch": "L",
    "cache_rate": 1,
    "cfg_scale": 1.0,
    "divisible": 16,
    "injection_args": {
      "injection_method": "concat",
      "replace_RGB": false
    },
    "patch_size": 2,
    "thermal_normalizer": 0.95941,
    "transport_config": {
      "loss_weight": "None",
      "path_type": "Linear",
      "prediction": "velocity"
    },
    "vae": "ema",
    "vae_config": {
      "act_fn": "silu",
      "block_out_channels": [
        128,
        256,
        512,
        512
      ],
      "down_block_types": [
        "DownEncoderBlock2D",
        "DownEncoderBlock2D",
        "DownEncoderBlock2D",
        "DownEncoderBlock2D"
      ],
      "force_upcast": true,
      "in_channels": 1,
      "latent_channels": 4,
      "layers_per_block": 2,
      "mid_block_add_attention": true,
      "norm_num_groups": 32,
      "out_channels": 1,
      "sample_size": 256,
      "up_block_types": [
        "UpDecoderBlock2D",
        "UpDecoderBlock2D",
        "UpDecoderBlock2D",
        "UpDecoderBlock2D"
      ],
      "use_post_quant_conv": true,
      "use_quant_conv": true
    },
    "vae_model": "klvae",
    "vae_path": "checkpoints/klvae_1st/checkpoints/last.ckpt"
  }
}