{
  "_class_name": "AutoencoderKLMiniMaxH3",
  "_diffusers_version": "0.40.0.dev0",
  "block_out_channels": [
    8,
    16
  ],
  "clip_length": 17,
  "decoder_attention_head_dim": 8,
  "decoder_ffn_mult": 2,
  "decoder_norm_eps": 1e-05,
  "decoder_num_attention_heads": 2,
  "decoder_num_layers": 2,
  "decoder_num_register_tokens": 2,
  "decoder_rope_dim_ratio": 0.75,
  "decoder_rope_theta": 100.0,
  "in_channels": 3,
  "latent_channels": 4,
  "latents_mean": [
    0.0,
    0.0,
    0.0,
    0.0
  ],
  "latents_std": [
    1.0,
    1.0,
    1.0,
    1.0
  ],
  "layers_per_block": 1,
  "norm_eps": 1e-06,
  "norm_num_groups": 8,
  "out_channels": 3,
  "spatial_downsample_factors": [
    4,
    4
  ],
  "spatial_padding_mode": "reflect",
  "temporal_downsample_factors": [
    2,
    2
  ],
  "token_drop": 3
}