{
  "_class_name": "Cosmos3OmniTransformer",
  "_diffusers_version": "0.39.0",
  "action_dim": 32,
  "action_gen": false,
  "attention_bias": false,
  "attention_dropout": 0.0,
  "backbone_type": "cosmos3_multiview",
  "base_fps": 30.0,
  "cosmos3_nano_sim_auto": {
    "attention_mode": "multiview",
    "cameras": [
      "camera_front_wide_120fov",
      "camera_cross_right_120fov",
      "camera_rear_right_70fov",
      "camera_rear_tele_30fov",
      "camera_rear_left_70fov",
      "camera_cross_left_120fov",
      "camera_front_tele_30fov",
      "camera_front_fisheye_200fov",
      "camera_left_fisheye_200fov",
      "camera_right_fisheye_200fov",
      "camera_rear_fisheye_200fov"
    ],
    "checkpoint_hash": "1c13c34424f1a040f9a7b4a73c469059a4347b50de730e7a7b7bbe8657577b49",
    "checkpoint_id": "sim_auto_joint_stage3_sfdmd_mads_all@iter_000001800",
    "checkpoint_iteration": 1800,
    "chunk": {
      "frames_per_chunk": 3
    },
    "fixed_step_sampler_config": {
      "num_train_timesteps": 1000,
      "sample_type": "sde",
      "t_list": [
        1.0,
        0.9375,
        0.8333333333333334,
        0.625
      ]
    },
    "inference_defaults": {
      "control_guidance": 1.0,
      "fps": 30.0,
      "guidance": 1.0,
      "resolution": "480"
    },
    "lidar": {
      "apply_validity_mask": true,
      "dtype": "float32",
      "fps": 10.0,
      "latent_channels": 128,
      "network_config": {
        "base_channels": 128,
        "bottleneck_3d": true,
        "bottleneck_3d_causal_time": true,
        "bottleneck_3d_max_t": 32,
        "bottleneck_3d_rope": true,
        "depths": [
          3,
          3,
          3,
          3
        ],
        "dilation": [
          1,
          1,
          1,
          1
        ],
        "formulation": "VAE",
        "in_channels": 3,
        "mapping_depth": 2,
        "mask_as_input": false,
        "mlp_ratio": 3.0,
        "num_heads": [
          4,
          4,
          8,
          8
        ],
        "out_channels": 3,
        "patch_size": [
          2,
          2
        ],
        "positional_embedding": "learnable_embedding",
        "resolution": [
          128,
          1808
        ],
        "temporal_downsample": [
          false,
          false,
          false
        ],
        "temporal_upsample": [
          false,
          false,
          false
        ],
        "window_size": [
          5,
          45
        ],
        "z_dim": 128
      },
      "range_projection": {
        "azimuth_end_degrees": -180.0,
        "azimuth_endpoint": false,
        "azimuth_start_degrees": 180.0,
        "coordinate_system": "x_forward_y_left_z_up",
        "intensity_encoding": "unit",
        "invalid_range_m": 0.0,
        "max_range_m": 100.0,
        "min_range_m": 5.0,
        "model_width": 1808,
        "model_width_transform": "circular_pad",
        "native_height": 128,
        "native_width": 3600,
        "return_selection": "nearest",
        "semantic_height": 128,
        "semantic_width": 1800,
        "sensor": "pandar128",
        "validity_threshold": 0.5
      },
      "sample_posterior": false,
      "spatial_compression": [
        16,
        16
      ],
      "streaming_chunk_frames": 20,
      "streaming_context_frames": 21,
      "temporal_compression_factor": 1
    },
    "lidar_latent_patch_size_hw": [
      1,
      1
    ],
    "mrope": {
      "base_fps": 30.0,
      "enable_fps_modulation": true,
      "reset_spatial_ids": true,
      "temporal_modality_margin": 15000
    },
    "prompt": {
      "context_after": 0,
      "context_before": 0,
      "emphasize_control_in_prompt": false,
      "fractional_duration": false,
      "media_rope_scale": 1.0,
      "media_text_extent": null,
      "policy": "segmented",
      "sink_age_cap_seconds": null,
      "system_prompt_joint": "av_joint_camera_lidar_transfer",
      "system_prompt_rgb": "av_multiview_transfer",
      "text_cache_max_len": 16384,
      "text_position_mode": "fixed",
      "text_sink_tokens": 4,
      "use_first_chunk_only": true
    },
    "replay": {
      "attention_scope": "decomposed",
      "clean_pass_causality": "chunk",
      "control_visibility": "causal",
      "controls_read_strict_past_clean_rgb": true,
      "cross_view_past_window_seconds": 0.4,
      "deduplicate_cross_view": true,
      "lidar_attends_captions": true
    },
    "rgb": {
      "latent_channels": 48,
      "latent_patch_size": 2,
      "temporal_compression_factor": 4,
      "vae_spatial_compression_factor": 16
    },
    "rig_view_embedding": {
      "camera_ids": {
        "camera_cross_left_120fov": 5,
        "camera_cross_right_120fov": 1,
        "camera_front_fisheye_200fov": 7,
        "camera_front_tele_30fov": 6,
        "camera_front_wide_120fov": 0,
        "camera_left_fisheye_200fov": 8,
        "camera_rear_fisheye_200fov": 10,
        "camera_rear_left_70fov": 4,
        "camera_rear_right_70fov": 2,
        "camera_rear_tele_30fov": 3,
        "camera_right_fisheye_200fov": 9
      },
      "lidar_id": 11,
      "num_embeddings": 12
    },
    "ring": {
      "cache_chunks": 2,
      "sink_chunks": 1
    },
    "schema_version": 1,
    "video_temporal_causal": true
  },
  "enable_fps_modulation": true,
  "head_dim": 128,
  "hidden_act": "silu",
  "hidden_size": 4096,
  "intermediate_size": 12288,
  "joint_attn_implementation": "multiview",
  "latent_channel": 48,
  "latent_patch_size": 2,
  "num_attention_heads": 32,
  "num_embodiment_domains": 32,
  "num_hidden_layers": 36,
  "num_key_value_heads": 8,
  "patch_latent_dim": 192,
  "qk_norm_for_text": true,
  "rms_norm_eps": 1e-06,
  "rope_axes_dim": [
    24,
    20,
    20
  ],
  "rope_scaling": {
    "mrope_interleaved": true,
    "mrope_section": [
      24,
      20,
      20
    ],
    "rope_type": "default"
  },
  "rope_theta": 5000000,
  "sound_dim": null,
  "sound_gen": false,
  "sound_latent_fps": 25,
  "temporal_compression_factor": 4,
  "timestep_scale": 0.001,
  "unified_3d_mrope_reset_spatial_ids": true,
  "unified_3d_mrope_temporal_modality_margin": 15000,
  "use_und_k_norm_for_gen": false,
  "video_temporal_causal": true,
  "vocab_size": 151936
}