wbhu / Tri-MipRF

Tri-MipRF: Tri-Mip Representation for Efficient Anti-Aliasing Neural Radiance Fields, ICCV'23 (Oral, Best Paper Finalist)

Home Page:https://wbhu.github.io/projects/Tri-MipRF

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

About the training resource

cskkxjk opened this issue · comments

Hi~ Thanks for your great work!
I wonder what kind of GPUs are required to train with the provided config config_files/ms_blender/TriMipRF.gin ?
I failed on 1080 due to some memory issues.

We tested it on A100-80g, but it only consumes about 9GB gpu memeory. If it still exceeds your gpu memory, please consider tune down Trainer.target_sample_batch_size

We tested it on A100-80g, but it only consumes about 9GB gpu memeory. If it still exceeds your gpu memory, please consider tune down Trainer.target_sample_batch_size

Thank you for the reply!

We tested it on A100-80g, but it only consumes about 9GB gpu memeory. If it still exceeds your gpu memory, please consider tune down Trainer.target_sample_batch_size

hi,I try to change the value of Trainer.target_sample_batch_size, change to large or small from 262144, but the GPU memory did not change, which parameter can control GPU-men?
thanks a lot!