Error running eval_3d_dense_caption on a model trained in huggingface

[1301358882]

may I ask that I plan to run eval_3d_dense_caption.sh with the trained model in huggingface, but there are always errors, as shown in the following figure, is it because my environment is different? My environment: python3.8.13, torch1.11.0+cu113

Thank you!

[ch3cook-fdu]

This might because of different versions of huggingface transformers.