[Error]"Specify the reduction dim with the 'dim' argument" occurs only with the text-xlarge model (includes temporary workaround)

When running example_text.py, I encounter the following error:
RuntimeError: max(): Expected reduction dim to be specified for input.numel() == 0. Specify the reduction dim with the 'dim' argument.

Even though this error appears, I can still run the code in the following way
Looking at the example code for the Text to 3D approach, we see:
pipeline = TrellisTextTo3DPipeline.from_pretrained("JeffreyXiang/TRELLIS-text-xlarge")

However, the above error only occurs with the text-xlarge model.
As a workaround, if I download the text-large or text-base models locally and then run:
TrellisTextTo3DPipeline.from_pretrained("/path/to/TRELLIS-text-large")
or
TrellisTextTo3DPipeline.from_pretrained("/path/to/TRELLIS-text-base")

the code works fine. I suspect that the text-xlarge model might be incompatible with certain environments.
Below is the result of running example_text.py with the text-large model.
This workaround is working well for me, so you might consider it as an option.

I'd appreciate it if you could provide a solution for resolving this error occurring with the text-xlarge model.

Start by reproducing the failure in example_text.py with TrellisTextTo3DPipeline.from_pretrained("JeffreyXiang/TRELLIS-text-xlarge"), then compare it with the text-large and text-base local models. Investigate the environment and model-loading differences visible from that entry point; done means the text-xlarge example runs without the reduction-dimension error.