search

malllabiisc / InteractE

AAAI 2020 - InteractE: Improving Convolution-based Knowledge Graph Embeddings by Increasing Feature Interactions
Apache License 2.0
89 stars 21 forks source link

How to set the parameters of reproducing WN18RR #2

Closed xjzhao001 closed 4 years ago

xjzhao001 commented 4 years ago

Hello, I am very interested in the method mentioned in your paper. Can you tell me all the parameters of reproducing WN18RR in detail? Looking forward to your reply, thank you!

soumyasanyal commented 4 years ago

Hi, we've mentioned the parameters for the different datasets in the README for the project. For WN18RR you can use this command - python interacte.py --data WN18RR --batch 256 --train_strategy one_to_n --feat_drop 0.2 --hid_drop 0.3 --perm 4 --ker_sz 11 --lr 0.001