Pytorch Roberta kernal died immediately when running " out = model(inputs)"

Question

I have a text dataset, which I trained on to get tokernizer, called "bert_tokenizer". Then I try to give a new word and get the word embedding out.

from transformers import RobertaConfig

config = RobertaConfig(
    vocab_enter code heresize=tokenizer.get_vocab_size(),
    max_position_embeddings=514,
    num_attention_heads=12,
    num_hidden_layers=6,
    type_vocab_size=1,)

#re-create tokenizer in transformers
from transformers import RobertaTokenizerFast

tokenizer = RobertaTokenizerFast.from_pretrained("bert_tokenizer", output_hidden_states =True, max_len=512)

#initialise model
from transformers import RobertaForMaskedLM

model = RobertaForMaskedLM(config=config)
model.eval()

word = tokenizer.encode('test test')
input = torch.LongTensor(word)
out = model(input_ids=input)

Failed the last line out = model(input_ids=input) , immediately. Error: kernel died. My training dataset is very small, is that a problem? Or other reasons?

I am following tutorial here: https://github.com/BramVanroy/bert-for-inference/blob/master/introduction-to-bert.ipynb

Thank you.