search

nlpyang / PreSumm

code for EMNLP 2019 paper Text Summarization with Pretrained Encoders
MIT License
1.29k stars 465 forks source link

visible gpu setting and bert model download #175

Open wailoktam opened 4 years ago

wailoktam commented 4 years ago

hi, I am sort of puzzled by the instruction on setting visible gpus to -1, kill the training process and rerun. Here are my questions:

  1. how do I know the download of bert is completed and I can stop the training session not using gpu?
  2. what happens if I just train with gpu = 1 from the very beginning?
  3. can I just download the bert model, put it somewhere for the program to find it and avoid the switch?

I am asking these questions because training with gpus = -1 and testing with gpus = 1 would result in error, not to mention the problem that training with gpus = -1 would take ages. Sticking with gpus = 1 seems more straightforward and I wonder how to do it.

SebastianVeile commented 4 years ago

This is only the case if you have more than 1 gpu. If you only use 1 gpu, try with gpus = 0. Then it will download the model and start training without any issues

Dejavu2019 commented 4 years ago

I have same questions with you, did you solve it?