Out of memory error

[Kirillspec]

I have 8Gb 2060 super nvidia card, windows 10 and miniconda, and i just installed pytorch with this command:
conda install pytorch=1.13.0 torchvision pytorch-cuda=11.6 -c pytorch -c nvidia
And i have "Out of memory" error on 5th step of Jupyter notebook running script "sample_text_to_3d".
Is there any possibilities to solve this problem?

[Kabanosk]

You can try lowering the batch_size parameter, eg.

batch_size = 1

[Kirillspec]

unfortunately it doesnt helps:
OutOfMemoryError: CUDA out of memory. Tried to allocate 768.00 MiB (GPU 0; 8.00 GiB total capacity; 6.14 GiB already allocated; 0 bytes free; 6.91 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

[MethodJiao]

Annotate or delete this code：
render_mode = 'nerf' # you can change this to 'stf'
size = 64 # this is the size of the renders; higher values take longer to render.
cameras = create_pan_cameras(size, device)
for i, latent in enumerate(latents):
images = decode_latent_images(xm, latent, cameras, rendering_mode=render_mode)
display(gif_widget(images))

[Kirillspec]

Annotate or delete this code： render_mode = 'nerf' # you can change this to 'stf' size = 64 # this is the size of the renders; higher values take longer to render. cameras = create_pan_cameras(size, device) for i, latent in enumerate(latents): images = decode_latent_images(xm, latent, cameras, rendering_mode=render_mode) display(gif_widget(images))

Thank you very much, it works!

[DenisIsDenis]

Hello! I have NVIDIA GeForce GTX 1650 with 4 GB of video memory.

I have this error:

torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 20.00 MiB (GPU 0; 4.00 GiB total capacity; 3.47 GiB already allocated; 0 bytes free; 3.47 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 have tried this:

1. decrease the value of max_split_size_mb;
2. reduce the value of batch_size to one`;
3. remove the code proposed by MethodJiao;

But none of the above helped solve the problem.

Just in case:

• Processor: HexaCore AMD Ryzen 5 3600, 3600 MHz (36 x 100)
• RAM: 32 GB
• OS: Windows 10

[Kabanosk]

I ran the model on google colab, and the memory needed to run it was around 7.6GB. I don't think there is an option now to reduce it to 4GB. If you want to try this model, this is link to my Google colab where I tested it