Attempt to reproduce BertExt model with PreSumm repo runs into error

[tanpengshi]

After following the instructions for data preparation such as the following:

python 5_make_data_files_presumm_mbert.py

I then proceed to the PreSumm repo to execute the instructions as given for model training (I was required to make some minor changes), but I ran into the following error:

return torch.embedding(weight, input, padding_idx, scale_grad_by_freq, sparse)

RuntimeError: index out of range at /pytorch/aten/src/TH/generic/THTensorEvenMoreMath.cpp:193

May I know if there was something like that that anyone has encountered? And how may I resolve the error?