Segmentation Fault while runing example from token classification

I am using the notebook provided in training token classification model. It runs fine until the trainer.Train() which causes jupyter kernel to crash. I tried using a python script to train and results in segmentation fault. I made some research online but see no solution. Can anyone help? Both cpu and gpu run into same issue
transformer version: 4.21.3
python: 3.7.5
dataset: 2.4.0
Please let me know if i should include other details as well. Thank you