search

nyu-systems / Grendel-GS

Ongoing research training gaussian splatting at scale by distributed system
Apache License 2.0
350 stars 19 forks source link

How to adapt to Gaussians with MLP #27

Open LaFeuilleMorte opened 3 weeks ago

LaFeuilleMorte commented 3 weeks ago

Hi, I was working on scaffold-gs and octree-gs which has a mlp for generating neural gaussians on the fly. How can I enable them to be trainable on in distributed way like the vanilla GS?

TarzanZhao commented 3 weeks ago

Sorry, we have not implemented distribution for these embedded neural networks. But you can simply have a DistributedDataParallel wrapper for such neural networks, and then you can train them distributed-ly.