getting error in train.py

[ankitkr3]

Hi @ramsrigouthamg

I am getting an error at the following line

tokenizer = T5Tokenizer.from_pretrained('t5-base')

error says : ImportError:
T5Tokenizer requires the SentencePiece library but it was not found in your environment. Checkout the instructions on the
installation page of its repo: https://github.com/google/sentencepiece#installation and follow the ones
that match your environment.

[aatithya-hitkar]

this worked for me
pip install transformers==2.9.0

[ankitkr3]

hi @aatithya-hitkar @ramsrigouthamg

I am getting segmentation fault as error while training on ec2 p2.xlarge instance. please help

[aatithya-hitkar]

i haven't faced this issue
but it may be because memory is not getting read
if you are working on multi gpu try single gpu

[ankitkr3]

@aatithya-hitkar @ramsrigouthamg
GPU issue it was thanks, but now it is getting trained continuously, it is currently showing epoch no. 14 while the param passed epoch is 2 only, can you help here?