search

jchibane / if-net

Implicit Feature Network (IF-Net) - Codebase
311 stars 57 forks source link

about the trained models #4

Closed littlesaohuo closed 4 years ago

littlesaohuo commented 4 years ago

Hi,

I really appreciate your great work. I am following the commands to prepare data and train, and it has taking a long time. Will you release the trained models? I want to learn the effect first.

Thanks!

nauman-tintash commented 4 years ago

Hi,

I was also trying to figure this out, we can probably afford to train the model once but if you're experimenting with different parameters, then it might not be feasible to do it over and over again.

One thing I tried while experimenting was to train on smaller datasets initially. For example, my use case was related to furniture so I only extracted and trained on the subset of Shapenet with data of Chairs.

But, eventually, it would be helpful if somebody posted a trained model. I will try to share mine but I haven't gotten around to training on the full dataset yet.

jchibane commented 4 years ago

Hi,

yes, I plan to release the models. However, I will prepare this in two weeks probably.

Best, Julian

Leerw commented 4 years ago

The pretrained model cannot be extracted as follows;

tar -xvf checkpoint_epoch_23.tar tar: This does not look like a tar archive tar: Skipping to next header tar: Exiting with failure status due to previous errors

jchibane commented 4 years ago

Hi,

the .tar file is already the checkpoint used by pytorch, i.e. you don't have to unzip it - but you can read it in directly. If I find some time in the future, I might add functionality for loading the pretrained models to the master branch. Other than that a PR is of course very welcome.

Best, Julian