config.json

{
  "accelerator": null,
  "action_dim": 128,
  "ema": {
    "inv_gamma": 1.0,
    "max_value": 0.9999,
    "min_value": 0.0,
    "power": 0.75,
    "update_after_step": 0
  },
  "enc_type": "theia-base-vit",
  "img_adaptor": "mlp2x_gelu",
  "img_cond_len": 4374,
  "img_pos_embed_config": [
    [
      "image",
      [
        2,
        3,
        -729
      ]
    ]
  ],
  "img_token_dim": 1152,
  "lang_adaptor": "mlp2x_gelu",
  "lang_pos_embed_config": [
    [
      "lang",
      -1024
    ]
  ],
  "lang_token_dim": 4096,
  "learnable_tokens": 196,
  "lora_config": null,
  "max_lang_cond_len": 1024,
  "noise_scheduler": {
    "beta_schedule": "squaredcos_cap_v2",
    "clip_sample": false,
    "num_inference_timesteps": 5,
    "num_train_timesteps": 1000,
    "prediction_type": "sample",
    "type": "ddpm"
  },
  "pred_horizon": 64,
  "rdt": {
    "cond_pos_embed_type": "multimodal",
    "depth": 28,
    "hidden_size": 2048,
    "num_heads": 32
  },
  "resolution": 256,
  "state_adaptor": "mlp3x_gelu",
  "state_token_dim": 128,
  "use_lora": false
}