config.json

{
  "architectures": [
    "Qwen3VLGeometryForConditionalGeneration"
  ],
  "depth_drop_prob": 0.0,
  "depth_loss_weight": 1.0,
  "depth_only_training": true,
  "dtype": "bfloat16",
  "eos_token_id": 151645,
  "geometry_config": {
    "dtype": "bfloat16",
    "enabled": true,
    "encoder_type": "lingbot_depth",
    "freeze_encoder": true,
    "fusion_layers": null,
    "fusion_method": "resize_and_add",
    "hidden_size": 1024,
    "match_post_merge_resolution": false,
    "merger_hidden_dim": null,
    "merger_type": "mlp",
    "model_name_or_path": null,
    "model_type": "geometry_encoder",
    "num_heads": 8,
    "num_layers": 1,
    "output_hidden_size": null,
    "reference_frame": "first",
    "use_3d_position_encoding": true,
    "encoder_model_config": {
      "encoder": {
        "backbone": "dinov2_vitl14",
        "intermediate_layers": 1,
        "dim_out": 1024,
        "strict": false,
        "depth_emb_mode": "conv_1c",
        "img_depth_fuse_mode": "cat_token"
      },
      "neck": {
        "dim_in": [
          1026,
          2,
          2,
          2,
          2
        ],
        "dim_out": null,
        "dim_res_blocks": [
          1024,
          256,
          128,
          64,
          32
        ],
        "num_res_blocks": [
          0,
          2,
          2,
          2,
          0
        ],
        "res_block_in_norm": "none",
        "res_block_hidden_norm": "none",
        "resamplers": [
          "conv_transpose",
          "conv_transpose",
          "conv_transpose",
          "bilinear"
        ]
      },
      "depth_head": {
        "dim_in": [
          1024,
          256,
          128,
          64,
          32
        ],
        "dim_out": [
          null,
          null,
          null,
          null,
          1
        ],
        "dim_res_blocks": [
          1024,
          256,
          128,
          64,
          32
        ],
        "num_res_blocks": [
          0,
          1,
          1,
          1,
          0
        ],
        "res_block_in_norm": "none",
        "res_block_hidden_norm": "none",
        "resamplers": [
          "conv_transpose",
          "conv_transpose",
          "conv_transpose",
          "bilinear"
        ]
      },
      "mask_head": {
        "dim_in": [
          1024,
          256,
          128,
          64,
          32
        ],
        "dim_out": [
          null,
          null,
          null,
          null,
          1
        ],
        "dim_res_blocks": [
          1024,
          256,
          128,
          64,
          32
        ],
        "num_res_blocks": [
          0,
          1,
          1,
          1,
          0
        ],
        "res_block_in_norm": "none",
        "res_block_hidden_norm": "none",
        "resamplers": [
          "conv_transpose",
          "conv_transpose",
          "conv_transpose",
          "bilinear"
        ]
      },
      "remap_output": "exp",
      "remap_depth_in": "log",
      "num_tokens_range": [
        1200,
        3600
      ]
    }
  },
  "image_token_id": 151655,
  "model_type": "qwen3_vl_geometry",
  "pad_token_id": 151643,
  "save_depth_viz_dir": null,
  "save_depth_viz_interval": 1000,
  "save_depth_viz_max_per_dataset": 20,
  "text_config": {
    "attention_bias": false,
    "attention_dropout": 0.0,
    "bos_token_id": 151643,
    "dtype": "bfloat16",
    "eos_token_id": 151645,
    "head_dim": 128,
    "hidden_act": "silu",
    "hidden_size": 4096,
    "initializer_range": 0.02,
    "intermediate_size": 12288,
    "max_position_embeddings": 262144,
    "model_type": "qwen3_vl_geometry_text",
    "num_attention_heads": 32,
    "num_hidden_layers": 36,
    "num_key_value_heads": 8,
    "rms_norm_eps": 1e-06,
    "rope_scaling": {
      "mrope_interleaved": true,
      "mrope_section": [
        24,
        20,
        20
      ],
      "rope_type": "default"
    },
    "rope_theta": 5000000,
    "use_cache": false,
    "vocab_size": 151936
  },
  "tie_word_embeddings": false,
  "transformers_version": "4.57.1",
  "use_cache": false,
  "use_depth_decoder": true,
  "use_encoder_output_for_depth_loss": false,
  "video_token_id": 151656,
  "vision_config": {
    "deepstack_visual_indexes": [
      8,
      16,
      24
    ],
    "depth": 27,
    "dtype": "bfloat16",
    "hidden_act": "gelu_pytorch_tanh",
    "hidden_size": 1152,
    "in_channels": 3,
    "initializer_range": 0.02,
    "intermediate_size": 4304,
    "model_type": "qwen3_vl_geometry",
    "num_heads": 16,
    "num_position_embeddings": 2304,
    "out_hidden_size": 4096,
    "patch_size": 16,
    "spatial_merge_size": 2,
    "temporal_patch_size": 2
  },
  "vision_end_token_id": 151653,
  "vision_start_token_id": 151652
}