About 'memory-optimized version' mentioned in paper

[Ruopeng Gao]

Hi, I've read your paper in ECCV 2022, believe that your research is truly meaningful.

I'm trying to reproduce your experiment. At section 4.2 in your paper, you have mentioned that "provide a memory-optimized version that can be trained on NVIDIA 2080 Ti GPUs". But I didn't find any details in this repository.

Did you release the memory-optimized code? If not, will you release this part?

Thanks a lot for your contribution.

[zyayoung]

Yes, checkpointing (memory optimization) is released.
Just add the --use_checkpoint argument (I think it is already added in the script).

[Ruopeng Gao]

It's my fault that misunderstood the meaning of the --use_checkpoint argument.
Thanks a lot!

[Bian-666]

It's my fault that misunderstood the meaning of the argument. Thanks a lot!--use_checkpoint

Hi，is the memory-optimized version trained on 8 2080ti or only single 2080ti?

[Ruopeng Gao]

It's my fault that misunderstood the meaning of the argument. Thanks a lot!--use_checkpoint

Hi，is the memory-optimized version trained on 8 2080ti or only single 2080ti?

I used this argument and trained on 8 TiTAN XP.

Using fewer cards for training will drop the performance.

[Ruopeng Gao]

Hi，is the memory-optimized version trained on 8 2080ti or only single 2080ti?

And if you use 2080Ti for training, 11GB CUDA Memory may not be enough. But you can change some code, then I think it will be fine on 2080Ti.

[Bian-666]

Hi，is the memory-optimized version trained on 8 2080ti or only single 2080ti?

And if you use 2080Ti for training, 11GB CUDA Memory may not be enough. But you can change some code, then I think it will be fine on 2080Ti.

thanks for your prompt reply，thanks a lot！

[Ruopeng Gao]

thanks for your prompt reply，thanks a lot！

😄 You're welcome.