config.json

{
  "chunk_size": 100,
  "dim_feedforward": 3200,
  "dim_model": 512,
  "dropout": 0.1,
  "feedforward_activation": "relu",
  "input_normalization_modes": {
    "observation.images.phone": "mean_std",
    "observation.images.web": "mean_std",
    "observation.state": "mean_std"
  },
  "input_shapes": {
    "observation.images.phone": [
      3,
      480,
      640
    ],
    "observation.images.web": [
      3,
      480,
      640
    ],
    "observation.state": [
      6
    ]
  },
  "kl_weight": 10.0,
  "latent_dim": 32,
  "n_action_steps": 100,
  "n_decoder_layers": 1,
  "n_encoder_layers": 4,
  "n_heads": 8,
  "n_obs_steps": 1,
  "n_vae_encoder_layers": 4,
  "output_normalization_modes": {
    "action": "mean_std"
  },
  "output_shapes": {
    "action": [
      6
    ]
  },
  "pre_norm": false,
  "pretrained_backbone_weights": "ResNet18_Weights.IMAGENET1K_V1",
  "replace_final_stride_with_dilation": false,
  "temporal_ensemble_coeff": null,
  "use_vae": true,
  "vision_backbone": "resnet18"
}