This repository has been archived by the owner on Nov 18, 2023. It is now read-only.
Pytorch Geometric support (deprecates KGCN) #161
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
jmsfltchr
changed the title
jmsfltchr
changed the title
added 25 commits
July 19, 2022 18:07
…uild due to tensorflow issue. A previously used patch is no longer applicable to v0.8.0 of rules_python.
…r, split out a separate labeller specifically for the binary link prediction case.
…or binary link prediction
…es (as in the diagnosis example)
…still throws an error
… came from the conversion of a homogeneous graph network into a heterogeneous one. Seemed impossible to debug the call graph creation in pytorch. Now the model runs, but the current issue is getting the model to see the negative samples too. Perhaps a different approach is required (NxN adjacency for the edge concerned).
…roach is easily applicable to predicting all edges in an adjacency matric, as per the link_prediction example.
…tures are given as tensors manually for each node. When given as numpy arrays this way of adding features works fine.
…layer, and now the learner learns with 100% accuracy on train and validation and 91% on test!
…ed for later use as well as the entry from edge_label_index_dict
added 8 commits
July 25, 2022 18:05
…se having an import conflict with the client package also called `typedb`.
jmsfltchr
force-pushed
the
pytorch-geometric-support
branch
from
bd30ab3 to
8ee8997
Compare
This was referenced Jul 27, 2022
Closed
added 9 commits
July 27, 2022 14:42
… removed query sampling capabilities which were poorly documented and never used.
…lines (it ignores patterns beginning `--`)
|
Approved for the build only. The API contains many changes and most of them are fundamental. They will be reviewed later in a longer discussion (most probably in person) with James and Joshua. |
lolski
approved these changes
Jul 28, 2022
| @@ -0,0 +1,66 @@ | |||
| # | |||
There was a problem hiding this comment.
What is the difference between requirements.txt and install_requires.txt?
There was a problem hiding this comment.
We discussed this in-person: install_requires.txt is used to fill the field of the same name for PyPi. We need this because the assemble-pip target doesn't support the flags that we need in the requirements.txt file. This is an issue and should be resolved better.
Closed
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What is the goal of this PR?
Migrate the ethos of TypeDB-ML to be a library to support machine learning using existing graph learning frameworks, starting with support for PyTorch Geometric and NetworkX to support in-memory graphs and general-purpose graph algorithms.
What are the changes implemented in this PR?
Solves #44