updated dgl dependency #268
Conversation
| else | ||
| pip3 uninstall dgl | ||
| pip3 install dgl-cu102==0.4.3post2 |
There was a problem hiding this comment.
Please add a comment about what is the latest dgl version that works with this patch. As we are using dgl._deprecate.graph.DGLGraph, this interface may be deleted in some date.
There was a problem hiding this comment.
Thank you @classicsong , I have added a minimum and maximum dependency for the tests
There was a problem hiding this comment.
This breaks the partition method (here: https://github.com/awslabs/dgl-ke/blob/master/python/dglke/partition.py#L119 ) because dgl switched the attribute name from dgl.transform to dgl.transforms in version 0.8.
There was a problem hiding this comment.
https://github.com/dmlc/dgl/releases/tag/0.8.0 under Graph Dataset and Transforms section
There was a problem hiding this comment.
I'm using version 0.9.1 of dgl and got this error. A reasonable work around for this would be to switch using the Version class in packaging.version.Version on version 0.8. I'll open a CR for this, (unless someone from this project indicates a preferred alternate way or says this shouldn't be done).
|
Hello @zheng-da @classicsong - |
tests/scripts/task_kg_test.sh
Outdated
| @@ -34,10 +34,10 @@ conda activate ${DGLBACKEND}-ci | |||
| # test | |||
| if [ "$2" == "cpu" ]; then | |||
| pip3 uninstall dgl | |||
| pip3 install dgl==0.4.3post2 | |||
| pip3 install dgl | |||
There was a problem hiding this comment.
why CPU version doesn't define the version number, but GPU version does?
There was a problem hiding this comment.
Thank you for catching this @zheng-da . My mistake - I added the version requirements to the wrong line - it should be correct now
Issue #, if available: #256
Description of changes:
dgl._deprecate.graph.DGLGraph. Effectively uses a deprecated class in a recent version of DGL instead of using a deprecated version of the DGL package.==0.4.3post2to>~0.5, <~0.9, allowing sharing a recent DGL package installation with other applications, and allowing use of Python binaries for 3.9+Integration tests have been updated and are running on my local environment.
I have re-ran training on FB15k to validate that the results are as expected, here are the output of the script:
The full training logs are available on this gist