IndexError: min(): Expected reduction dim 0 to have non-zero size.

When I run demo.py, I meet problem below. Can u give me a clue?
```
❯ python demo.py
/root/miniconda3/envs/sam3d-objects-lc/lib/python3.11/site-packages/torch/cuda/__init__.py:61: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml directly, please report this to the maintainers of the package that installed pynvml for you.
import pynvml # type: ignore[import]
Warp 1.10.0 initialized:
CUDA Toolkit 12.8, Driver 12.4
Devices:
"cpu" : "x86_64"
"cuda:0" : "NVIDIA A100-SXM4-80GB" (79 GiB, sm_80, mempool enabled)
"cuda:1" : "NVIDIA A100-SXM4-80GB" (79 GiB, sm_80, mempool enabled)
"cuda:2" : "NVIDIA A100-SXM4-80GB" (79 GiB, sm_80, mempool enabled)
"cuda:3" : "NVIDIA A100-SXM4-80GB" (79 GiB, sm_80, mempool enabled)
"cuda:4" : "NVIDIA A100-SXM4-80GB" (79 GiB, sm_80, mempool enabled)
"cuda:5" : "NVIDIA A100-SXM4-80GB" (79 GiB, sm_80, mempool enabled)
"cuda:6" : "NVIDIA A100-SXM4-80GB" (79 GiB, sm_80, mempool enabled)
"cuda:7" : "NVIDIA A100-SXM4-80GB" (79 GiB, sm_80, mempool enabled)
CUDA peer access:
Supported fully (all-directional)
Kernel cache:
/root/.cache/warp/1.10.0
2025-11-27 17:02:17.731 | INFO | sam3d_objects.pipeline.inference_pipeline:set_attention_backend:15 - GPU name is NVIDIA A100-SXM4-80GB
2025-11-27 17:02:19.689 | INFO | sam3d_objects.model.backbone.tdfy_dit.modules.sparse:__from_env:39 - [SPARSE] Backend: spconv, Attention: flash_attn
2025-11-27 17:02:24.415 | INFO | sam3d_objects.model.backbone.tdfy_dit.modules.attention:__from_env:30 - [ATTENTION] Using backend: flash_attn
[SPARSE][CONV] spconv algo: auto
2025-11-27 17:02:24.765 | WARNING | sam3d_objects.data.dataset.tdfy.preprocessor:__post_init__:51 - No rgb pointmap normalizer provided, using scale + shift
2025-11-27 17:02:24.766 | WARNING | sam3d_objects.data.dataset.tdfy.preprocessor:__post_init__:51 - No rgb pointmap normalizer provided, using scale + shift
2025-11-27 17:02:24.807 | WARNING | sam3d_objects.data.dataset.tdfy.preprocessor:__post_init__:51 - No rgb pointmap normalizer provided, using scale + shift
model.pt: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1.26G/1.26G [00:53<00:00, 23.6MB/s]
/root/miniconda3/envs/sam3d-objects-lc/lib/python3.11/site-packages/moge/model/v1.py:171: UserWarning: The following deprecated/invalid arguments are ignored: {'output_mask': True, 'split_head': True}
warnings.warn(f"The following deprecated/invalid arguments are ignored: {deprecated_kwargs}")
2025-11-27 17:03:24.900 | WARNING | sam3d_objects.data.dataset.tdfy.preprocessor:__post_init__:51 - No rgb pointmap normalizer provided, using scale + shift
2025-11-27 17:03:24.901 | INFO | sam3d_objects.pipeline.inference_pipeline:__init__:98 - self.device: cuda
2025-11-27 17:03:24.901 | INFO | sam3d_objects.pipeline.inference_pipeline:__init__:99 - CUDA_VISIBLE_DEVICES: None
2025-11-27 17:03:24.901 | INFO | sam3d_objects.pipeline.inference_pipeline:__init__:100 - Actually using GPU: 0
2025-11-27 17:03:24.901 | INFO | sam3d_objects.pipeline.inference_pipeline:init_pose_decoder:295 - Using pose decoder: ScaleShiftInvariant
2025-11-27 17:03:24.901 | INFO | sam3d_objects.pipeline.inference_pipeline:__init__:131 - Loading model weights...
2025-11-27 17:03:25.112 | INFO | sam3d_objects.model.io:load_model_from_checkpoint:158 - Loading checkpoint from checkpoints/hf/ss_generator.ckpt
2025-11-27 17:03:32.935 | INFO | sam3d_objects.model.io:load_model_from_checkpoint:158 - Loading checkpoint from checkpoints/hf/slat_generator.ckpt
2025-11-27 17:03:37.443 | INFO | sam3d_objects.model.io:load_model_from_checkpoint:158 - Loading checkpoint from checkpoints/hf/ss_decoder.ckpt
2025-11-27 17:03:37.851 | INFO | sam3d_objects.model.io:load_model_from_checkpoint:158 - Loading checkpoint from checkpoints/hf/slat_decoder_gs.ckpt
2025-11-27 17:03:38.216 | INFO | sam3d_objects.model.io:load_model_from_checkpoint:158 - Loading checkpoint from checkpoints/hf/slat_decoder_gs_4.ckpt
2025-11-27 17:03:38.605 | INFO | sam3d_objects.model.io:load_model_from_checkpoint:158 - Loading checkpoint from checkpoints/hf/slat_decoder_mesh.ckpt
2025-11-27 17:03:39.100 | INFO | sam3d_objects.model.backbone.dit.embedder.dino:__init__:31 - Loading DINO model: dinov2_vitl14_reg from facebookresearch/dinov2 (source: github)
2025-11-27 17:03:41.387 | INFO | sam3d_objects.model.backbone.dit.embedder.dino:__init__:44 - Loaded DINO model - type: , embed_dim: 1024, patch_size: (14, 14)
2025-11-27 17:03:41.395 | INFO | sam3d_objects.model.backbone.dit.embedder.dino:__init__:31 - Loading DINO model: dinov2_vitl14_reg from facebookresearch/dinov2 (source: github)
2025-11-27 17:03:43.441 | INFO | sam3d_objects.model.backbone.dit.embedder.dino:__init__:44 - Loaded DINO model - type: , embed_dim: 1024, patch_size: (14, 14)
2025-11-27 17:03:43.556 | INFO | sam3d_objects.model.io:load_model_from_checkpoint:158 - Loading checkpoint from checkpoints/hf/ss_generator.ckpt
2025-11-27 17:03:48.874 | INFO | sam3d_objects.model.backbone.dit.embedder.dino:__init__:31 - Loading DINO model: dinov2_vitl14_reg from facebookresearch/dinov2 (source: github)
2025-11-27 17:03:50.373 | INFO | sam3d_objects.model.backbone.dit.embedder.dino:__init__:44 - Loaded DINO model - type: , embed_dim: 1024, patch_size: (14, 14)
2025-11-27 17:03:50.378 | INFO | sam3d_objects.model.backbone.dit.embedder.dino:__init__:31 - Loading DINO model: dinov2_vitl14_reg from facebookresearch/dinov2 (source: github)
2025-11-27 17:03:52.207 | INFO | sam3d_objects.model.backbone.dit.embedder.dino:__init__:44 - Loaded DINO model - type: , embed_dim: 1024, patch_size: (14, 14)
2025-11-27 17:03:52.219 | INFO | sam3d_objects.model.io:load_model_from_checkpoint:158 - Loading checkpoint from checkpoints/hf/slat_generator.ckpt
2025-11-27 17:03:56.257 | INFO | sam3d_objects.pipeline.inference_pipeline:override_ss_generator_cfg_config:436 - ss_generator parameters: inference_steps=25, cfg_strength=7, cfg_interval=[0, 500], rescale_t=3, cfg_strength_pm=0.0
2025-11-27 17:03:56.258 | INFO | sam3d_objects.pipeline.inference_pipeline:override_slat_generator_cfg_config:458 - slat_generator parameters: inference_steps=25, cfg_strength=1, cfg_interval=[0, 500], rescale_t=1
2025-11-27 17:03:56.258 | INFO | sam3d_objects.pipeline.inference_pipeline:__init__:196 - Loading model weights completed!
2025-11-27 17:04:00.579 | INFO | sam3d_objects.pipeline.inference_pipeline:sample_sparse_structure:662 - Sampling sparse structure: inference_steps=25, strength=7, interval=[0, 500], rescale_t=3, cfg_strength_pm=0.0
2025-11-27 17:04:00.579 | INFO | sam3d_objects.pipeline.inference_pipeline:get_condition_input:631 - Running condition embedder ...
2025-11-27 17:04:00.755 | INFO | sam3d_objects.pipeline.inference_pipeline:get_condition_input:635 - Condition embedder finishes!
Traceback (most recent call last):
File "/mnt/data/lc/sam-3d-objects/demo.py", line 17, in
output = inference(image, mask, seed=42)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/mnt/data/lc/sam-3d-objects/notebook/inference.py", line 109, in __call__
return self._pipeline.run(
^^^^^^^^^^^^^^^^^^^
File "/mnt/data/lc/sam-3d-objects/sam3d_objects/pipeline/inference_pipeline_pointmap.py", line 351, in run
ss_return_dict = self.sample_sparse_structure(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/mnt/data/lc/sam-3d-objects/sam3d_objects/pipeline/inference_pipeline.py", line 707, in sample_sparse_structure
coords = prune_sparse_structure(
^^^^^^^^^^^^^^^^^^^^^^^
File "/mnt/data/lc/sam-3d-objects/sam3d_objects/pipeline/inference_utils.py", line 374, in prune_sparse_structure
min_xyz = coords.min(0)[0]
^^^^^^^^^^^^^
File "/root/miniconda3/envs/sam3d-objects-lc/lib/python3.11/site-packages/torch/utils/_device.py", line 106, in __torch_function__
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
IndexError: min(): Expected reduction dim 0 to have non-zero size.
```