search

icon-lab / SynDiff

Official PyTorch implementation of SynDiff described in the paper (https://arxiv.org/abs/2207.08208).
Other
207 stars 38 forks source link

CUDA out of memory issue #30

Closed hanajibsa closed 7 months ago

hanajibsa commented 8 months ago

During training, CUDA out of memory emerged. I think there are something that accumulate memory, but I can't find.

muzafferozbey commented 7 months ago

Thank you for your interest in our work and for raising this issue. Our models are optimized for Nvidia RTX3090 (24 GB) but were initially intended to work with 16GB GPUs. If you're encountering out-of-memory errors, consider adjusting the model complexity by modifying the ch_mult or num_res_blocks configurations.

We trust that this codebase will prove valuable for your research endeavors. If you have any other questions or concerns, please feel free to ask.

guoliangq commented 5 months ago

I also think there are something that accumulate memory, but I can't find. please check it.