config.json

{
  "n_obs_steps": 5,
  "n_action_pred_token": 7,
  "n_action_pred_chunk": 5,
  "input_shapes": {
    "observation.image": [
      3,
      96,
      96
    ],
    "observation.state": [
      2
    ]
  },
  "output_shapes": {
    "action": [
      2
    ]
  },
  "input_normalization_modes": {
    "observation.image": "mean_std",
    "observation.state": "min_max"
  },
  "output_normalization_modes": {
    "action": "min_max"
  },
  "vision_backbone": "resnet18",
  "crop_shape": [
    84,
    84
  ],
  "crop_is_random": true,
  "pretrained_backbone_weights": null,
  "use_group_norm": true,
  "spatial_softmax_num_keypoints": 32,
  "discretize_step": 20000,
  "vqvae_groups": 2,
  "vqvae_n_embed": 16,
  "vqvae_embedding_dim": 256,
  "gpt_block_size": 500,
  "gpt_output_dim": 512,
  "gpt_n_layer": 8,
  "gpt_n_head": 8,
  "gpt_hidden_dim": 512,
  "dropout": 0.1,
  "mlp_hidden_dim": 1024,
  "offset_loss_weight": 10000.0
}