{
  "architectures": [
    "VAEModel"
  ],
  "attn_resolutions": [],
  "auto_map": {
    "AutoConfig": "configuration_vae.VAEConfig",
    "AutoModel": "modeling_vae.VAEModel"
  },
  "channels": 128,
  "channels_mult": [
    1,
    1,
    1,
    2,
    2
  ],
  "codebook_dim": 0,
  "codebook_size": 0,
  "decoder_type": "Simple",
  "drop_out": 0,
  "dropout": 0.0,
  "encoder_type": "Simple",
  "image_mean": [
    0.5,
    0.5,
    0.5
  ],
  "image_std": [
    0.5,
    0.5,
    0.5
  ],
  "in_channels": 3,
  "model_type": "vae",
  "num_res_blocks": 2,
  "out_channels": 3,
  "quantizer_type": "VQ",
  "resolution": [
    64,
    64
  ],
  "torch_dtype": "float32",
  "transformers_version": "4.51.3",
  "w_commit": 0,
  "w_dino": 0,
  "w_kl": 1,
  "w_l1": 0.2,
  "w_mse": 2,
  "w_perceptual": 0,
  "z_channels": 64
}