search

eladrich / latent-nerf

Official Implementation for "Latent-NeRF for Shape-Guided Generation of 3D Shapes and Textures"
MIT License
687 stars 48 forks source link

CUDA out of memory issue #27

Open mshooter opened 1 year ago

mshooter commented 1 year ago

Hi, I keep getting Cuda out of memory... Which parameters should I change for this?

flow-specter commented 1 year ago

I have the same issue...

qsong2001 commented 1 year ago

if you met this issue: RuntimeError: CUDA error: an illegal memory access was encountered

You can simply try to re-run the training script repeatedly (like 4,5 times) . It worked for me.

SzaremehrjardiMT commented 4 months ago

Same issue on 4GB RAM. torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 26.00 MiB (GPU 0; 4.00 GiB total capacity; 3.38 GiB already allocated; 0 bytes free; 3.42 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF I tried different solutions such as reducing batch size, using torch.no_grad() for Inference and loading models in float16 precision instead of the default float32 precision; But none of them worked for me.

What is the minimum hardware requirements?