Skip to content

Commit

Permalink
update docker (dmlc#1571)
Browse files Browse the repository at this point in the history
  • Loading branch information
barry-jin authored May 18, 2021
1 parent 5122e84 commit df52171
Show file tree
Hide file tree
Showing 2 changed files with 3 additions and 7 deletions.
2 changes: 1 addition & 1 deletion tools/batch/submit-job.py
Original file line number Diff line number Diff line change
Expand Up @@ -10,7 +10,7 @@

instance_type_info = {
'g4dn.4x': {
'job_definition': 'gluon-nlp-g4dn_4xlarge:5',
'job_definition': 'gluon-nlp-g4dn_4xlarge:6',
'job_queue': 'g4dn'
},
'g4dn.8x': {
Expand Down
8 changes: 2 additions & 6 deletions tools/docker/gluon_nlp_job.sh
Original file line number Diff line number Diff line change
Expand Up @@ -23,14 +23,10 @@ git fetch origin $SOURCE_REF:working
git checkout working

if [ $DEVICE == "cpu" ]; then
python3 -m pip uninstall --quiet mxnet -y
python3 -m pip install -U --quiet --pre "mxnet>=2.0.0b20210121" --user
else
# Due to the issue in https://forums.aws.amazon.com/thread.jspa?messageID=953912
# We need to manually configure the shm to ensure that Horovod is runnable.
# The reason that we need a larger shm is described in https://github.com/NVIDIA/nccl/issues/290
umount shm
mount -t tmpfs -o rw,nosuid,nodev,noexec,relatime,size=2G shm /dev/shm
sudo python3 -m pip uninstall --quiet mxnet-cu102
python3 -m pip uninstall --quiet mxnet-cu102 -y
python3 -m pip install -U --quiet --pre "mxnet-cu102>=2.0.0a" --user
fi

Expand Down

0 comments on commit df52171

Please sign in to comment.