DeepRec-AI/DeepRec

How to use workqueue?

zhoujingzhe opened this issue · 1 comments

using modelzoo/simple_multitask as an example.

If launch a training program with "python train.py", there are lots of logs like "loss=0.606937, steps=1".
However, if launch a training program with "python train.py --workqueue=True", there are only one info "loss=0.606937, steps=1".

In the logimage
why restore work queue twice? Especially after training completed?

good catch, there's bug in the modelzoo/simple_multitask. We have fix it with the PR: #839