{
  "_class_name": "AutoencoderKL",
  "_diffusers_version": "0.35.1",
  "_use_default_values": [
    "use_post_quant_conv",
    "latents_std",
    "mid_block_add_attention",
    "use_quant_conv",
    "shift_factor",
    "latents_mean"
  ],
  "act_fn": "silu",
  "block_out_channels": [
    128,
    256,
    512,
    512
  ],
  "down_block_types": [
    "DownEncoderBlock2D",
    "DownEncoderBlock2D",
    "DownEncoderBlock2D",
    "DownEncoderBlock2D"
  ],
  "force_upcast": true,
  "in_channels": 3,
  "latent_channels": 4,
  "latents_mean": null,
  "latents_std": null,
  "layers_per_block": 2,
  "mid_block_add_attention": true,
  "neuron": {
    "auto_cast": "matmul",
    "auto_cast_type": "bf16",
    "compiler_type": "neuronx-cc",
    "compiler_version": "2.19.8089.0+8ab9f450",
    "dynamic_batch_size": false,
    "float_dtype": "fp32",
    "inline_weights_to_neff": true,
    "input_names": [
      "sample"
    ],
    "int_dtype": "int64",
    "model_type": "vae-encoder",
    "optlevel": "2",
    "output_attentions": false,
    "output_hidden_states": false,
    "output_names": [
      "latent_parameters"
    ],
    "static_batch_size": 1,
    "static_height": 1024,
    "static_num_channels": 3,
    "static_width": 1024,
    "task": "semantic-segmentation",
    "tensor_parallel_size": 1
  },
  "norm_num_groups": 32,
  "out_channels": 3,
  "output_attentions": false,
  "sample_size": 1024,
  "scaling_factor": 0.13025,
  "shift_factor": null,
  "transformers_version": null,
  "up_block_types": [
    "UpDecoderBlock2D",
    "UpDecoderBlock2D",
    "UpDecoderBlock2D",
    "UpDecoderBlock2D"
  ],
  "use_post_quant_conv": true,
  "use_quant_conv": true
}