Skip to content
This repository has been archived by the owner on Jul 10, 2024. It is now read-only.

SUBMARINE-561. [SDK] Add PyTorch implementation of AFM model #346

Closed
wants to merge 26 commits into from

Conversation

ifndef012
Copy link
Contributor

What is this PR for?

Add PyTorch implementation of Attentional Factorization Machine for CTR prediction. (AFM)
Make minor modifications to the PyTorch training flow.
Add testing for the AFM model.

What type of PR is it?

[Improvement]

Todos

  • - Task

What is the Jira issue?

https://issues.apache.org/jira/browse/SUBMARINE-561

How should this be tested?

python-sdk
Submarine

Screenshots (if appropriate)

Questions:

  • Does the licenses files need update? No
  • Is there breaking changes for older versions? No
  • Does this needs documentation? No

Copy link
Member

@pingsutw pingsutw left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks @andrewhsiehth for the contribution
Overall LGTM, some suggestions.

  1. Could we rename AttentionalFM to AFM, it's easier to use for users.
  2. We could add deepfm and afm directories in submarine-sdk/pysubmarine/example/pytorch/, and put the related file to the directories. After refactor, make sure to update the train_data path to ../../data/tr.libsvm

Tensorflow deepfm example:
https://github.com/apache/submarine/tree/master/submarine-sdk/pysubmarine/example/tensorflow

@pingsutw
Copy link
Member

+1, Thanks @andrewhsiehth
Will commit if no more comment.

@ifndef012
Copy link
Contributor Author

@pingsutw Thanks for your suggestions and review.

Copy link
Member

@xunliu xunliu left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@andrewhsiehth Thank you contribution this feature. :-)

@asfgit asfgit closed this in e580802 Jul 27, 2020
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

3 participants