search

anonymous-dense-submission / DensE

MIT License
4 stars 3 forks source link

Pretrained models #1

Open Demirrr opened 3 years ago

Demirrr commented 3 years ago

Hello all,

I was wondering whether the pretrained models will be publicly available.

Cheers

luhaonan20 commented 3 years ago

Hi,

In principle, the results can be reproduced within the statistical error easily by using the same hyperparameter setting as we mentioned here. We trained the model based on V100 GPU in our lab, and we can also publish the pre-trained models somewhere if you need it 👍

Cheers

Demirrr commented 3 years ago

Dear @luhaonan20,

thank you for your prompt reply.

I reckon publishing the pre-trained models somewhere would not only facilitate the research in this direction but also be helpful for the mother nature due to energy consumption :) As you might very well known without V100 GPU the computation of embeddings on WN18RR, FB15k-237 and YAGO3-10 would take couple of days on a commodity computer.

I would personally be very grateful if you could provide the pre-trained models.

Cheers

aalexandrasilva commented 3 years ago

Dear @luhaonan20,

Would it be possible to publish the pre-trained models as you've previously mentioned in this issue? Or in case they have been published already, could you please point me to their location?

Thank you in advance.