torch.OutOfMemoryError: CUDA out of memory.

[reset12]

I am not be able to use CUDA on my Acer Nitro V15 with Nvidia Geforce RTX Card.

I get the following error
"torch.OutOfMemoryError: CUDA out of memory.
Tried to allocate 1024.00 MiB. GPU 0 has a
total capacity of 5.64 GiB of which 798.50 MiB is free.
Including non-PyTorch memory, this process has 4.83 GiB memory in use.
Of the allocated memory 4.60 GiB is allocated by PyTorch, and 129.54 MiB
is reserved by PyTorch but unallocated.
If reserved but unallocated memory is large try setting
PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation."

I have tried different configurations without any success.
"cat Meshroom.sh
#!/bin/sh
#(export 'PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True';/home/data/Meshroom-2025.1.0/Meshroom)
#(export 'PYTORCH_CUDA_ALLOC_CONF=max_split_size_mb:128';/home/data/Meshroom-2025.1.0/Meshroom)
(export 'PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True,max_split_size_mb:750';/home/data/Meshroom-2025.1.0/Meshroom)
"

Desktop (please complete the following and other pertinent information):

• OS: xubuntu 26.04
• Python 3.13.11
• Qt/PySide version [e.g. 6.8.2]
• Meshroom version: 2025.1.0
• nvidia-graphics-drivers-580

When i switch of GPU on ImageSegmentationBox it is running, but i have buyed extra a new Laptop with Nvidia GPU, so i like to use it.