-
Notifications
You must be signed in to change notification settings - Fork 1.2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Pin mxnet version in response to mx.metric reorg #1310
Conversation
Job PR-1310-1 is done. |
@dmlc/gluon-nlp-committers |
gluoncv/__init__.py
Outdated
# pylint: disable=unused-import | ||
from mxnet import metric | ||
except ImportError: | ||
from mxnet.gluon import metric |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
There are API changes besides the rename: apache/mxnet#18046
@leezu What's nlp's solution in terms of the transition from 1.0 to 2.0 at this time? |
@zhreshold Currently, we are improving lots of APIs. The new numpy version will rely on MXNet 2.0 only and will not be compatible to the old version: dmlc/gluon-nlp#1225 |
@sxjscience So you will pin the mxnet to be <2.0 for non-numpy gluonnlp? |
@zhreshold Yes, I think it will depend on MXNet 1.7. Also @dmlc/gluon-nlp-committers , please correct me if I'm wrong. |
@dmlc/gluon-cv-committers @dmlc/gluon-cv-reviewer GluonCV hasn't started transition to either numpy or mxnet 2.0 so the choice will be
|
the cost of committing to backward compatibility for 2.0 is too high and infeasible. I think proper versioning should be used to freeze gluoncv to 1.x until the transition happens. |
Agree with szha@ and sxjscience@, it is easier to just pin the version to 1.x before the transition is fully ready. |
Job PR-1310-5 is done. |
* mx.metric * pin version * fix lint * reduce parallel test iter/bs
FYI mx.metric is replaced by mx.gluon.metric in mxnet 2.0