Replies: 1 comment 1 reply
-
The relation matching (RM) module is the key. If you include that and run for larger epochs, I think you would get better results. RM module can be found in the original paper's repository. I hope that this clears your doubt. Also, a hyperparameter search could also help to filter out the useful ones. |
Beta Was this translation helpful? Give feedback.
1 reply
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
Hi,
I get 0.68 in metaQA half KG-hop1. I'm using your command with the same hyperparameter setting. After the first epoch, I get 0.68, and then the score stays the same. After 10 epochs, the training automatically ends.0.68 is too low, I do not know what's wrong with it? I would like to ask whether you get the result 83.9 in the paper?
Beta Was this translation helpful? Give feedback.
All reactions