how to run it? about embedkgqa HOT 3 CLOSED

Knowledge graph embedding:

1. Download dataset from KnowledgeGraphEmbedding.git. This project contain almost dataset we need for KGE.
2. Enter directory train_embeddings, activate python3 env. run command with some arguments in shell:

python3 main.py \
--dataset FB15k \
--model Rotat3 \
--cuda True

note : create directory kg_embeddings in EmbedKBQA before trainning.

KBQA:

1. Download RoBERTa into ./KGQA, in case of long downloading cost.
2. Download MetaQA dataset, and WebQuestionSP dataset.
3. modify some directory in ./KGQA/main.py, specify dataset path.
4. run ./KGQA/main.py.
   BTW, I am not run this command success yet. Since I cost lots of time in downloading RoBERTa.

from embedkgqa.

Hi, thanks for your comments @HTMLgtMK . Some of the instructions you listed might be very useful!
The complete instructions/dataset are yet to be posted. Unfortunately I'm busy with some stuff on my end which is causing this delay. Practically, please expect this to take another week or two to get a properly documented solution for reproducing the results stated in our paper.

The problem is that the current working code/data is not organized properly and needs a bit of refactoring to make it run easily on other machines. The currently uploaded code, however, is the core of it.

I hope this is ok. In any case, if this delay seems unjustified, please do let me know and I'll see what I can do.
I'll be keeping this issue pinned in case anyone else also wants to comment on this.
Thanks!

from embedkgqa.

Added instructions in the README and updated the code, please check @lbda1

from embedkgqa.