DrRepair icon indicating copy to clipboard operation
DrRepair copied to clipboard

Published 20 hours ago • verified publisher icon michiyasunaga

0% trained data trained

Open akash-isu opened this issue 4 years ago • 4 comments

I keep following the steps in the ReadMe but whatever I do after the training phase it always shows TRAIN: 0% |.

akash-isu avatar Aug 06 '20 16:08 akash-isu

Hi, perhaps this is a GPU issue - could you double check if your GPU is properly configured and recognized by pytorch?

michiyasunaga avatar Aug 27 '20 22:08 michiyasunaga

I do not have CUDA on my system. I tried removing the cuda references in the codebase. However, it still doesn't work.

akash-isu avatar Aug 28 '20 16:08 akash-isu

A small update. If I go ahead and remove the references to CUDA in the err_localize_edit.py and then run the code, I get the list index out of range error. Not sure how these are related.

akash-isu avatar Aug 29 '20 21:08 akash-isu

You can reduce the no. of steps in DrRepair/model/configs/data-spoc/err-data-orig.yml

timing:
  max_steps: 150000

vj68 avatar May 19 '21 15:05 vj68