search

yenchenlin / nerf-pytorch

A PyTorch implementation of NeRF (Neural Radiance Fields) that reproduces the results.
MIT License
5.33k stars 1.04k forks source link

Problems with running pre trained models #133

Open zfkdsg opened 6 months ago

zfkdsg commented 6 months ago

I want to observe the effect on my own computer, and I downloaded the logo from pre train. However, I encountered a bug while running the sentence "run nerf. py -- config configs/lego. txt -- rended_only" in Python ：
torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 2.11 GiB. GPU 0 has a total capacity of 3.81 GiB of which 664.00 MiB is free. Including non-PyTorch memory, this process has 3.15 GiB memory in use. Of the allocated memory 2.41 GiB is allocated by PyTorch, and 654.21 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. See documentation for Memory Management (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)

please help me thanks ！！！

GauravNerf commented 5 months ago

just delete the pycache folder before running the code that might be pre generated from the failed previous training and then restart the render_only command, this should work unless you have very low gpu memory.