Tensor.item() cannot be called on meta tensors

When I run app.py, this error occurs. What should I do?

File "/projects/github_project/TRELLIS.2/app.py", line 633, in
pipeline = Trellis2ImageTo3DPipeline.from_pretrained('/projects/github_project/TRELLIS.2/models/TRELLIS.2-4B')
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/projects/github_project/TRELLIS.2/trellis2/pipelines/trellis2_image_to_3d.py", line 105, in from_pretrained
pipeline.rembg_model = getattr(rembg, args['rembg_model']['name'])(**args['rembg_model']['args'])
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/projects/github_project/TRELLIS.2/trellis2/pipelines/rembg/BiRefNet.py", line 11, in init
self.model = AutoModelForImageSegmentation.from_pretrained(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/projects/github_project/TRELLIS.2/.venv/lib/python3.12/site-packages/transformers/models/auto/auto_factory.py", line 365, in from_pretrained
return model_class.from_pretrained(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/projects/github_project/TRELLIS.2/.venv/lib/python3.12/site-packages/transformers/modeling_utils.py", line 4072, in from_pretrained
model = cls(config, *model_args, **model_kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/jiuzhang/.cache/huggingface/modules/transformers_modules/briaai/birefnet.py", line 1986, in init
self.bb = build_backbone(self.config.bb, pretrained=bb_pretrained)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/jiuzhang/.cache/huggingface/modules/transformers_modules/briaai/birefnet.py", line 1405, in build_backbone
bb = eval('{}({})'.format(bb_name, params_settings))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "", line 1, in
File "/home/jiuzhang/.cache/huggingface/modules/transformers_modules/briaai/birefnet.py", line 1224, in swin_v1_l
model = SwinTransformer(embed_dim=192, depths=[2, 2, 18, 2], num_heads=[6, 12, 24, 48], window_size=12)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/jiuzhang/.cache/huggingface/modules/transformers_modules/briaai/birefnet.py", line 1131, in init
dpr = [x.item() for x in torch.linspace(0, drop_path_rate, sum(depths))] # stochastic depth decay rule
^^^^^^^^
File "/projects/github_project/TRELLIS.2/.venv/lib/python3.12/site-packages/torch/utils/_device.py", line 104, in torch_function
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "/projects/github_project/TRELLIS.2/.venv/lib/python3.12/site-packages/torch/_meta_registrations.py", line 6585, in meta_local_scalar_dense
raise RuntimeError("Tensor.item() cannot be called on meta tensors")
RuntimeError: Tensor.item() cannot be called on meta tensors

Start at app.py line 633 and trace Trellis2ImageTo3DPipeline.from_pretrained into trellis2/pipelines/trellis2_image_to_3d.py and trellis2/pipelines/rembg/BiRefNet.py. Reproduce the BiRefNet loading failure around the reported Tensor.item() call and determine the project-supported resolution; done means pipeline initialization completes without the meta-tensor error.