search

ymxlzgy / echoscene

[ECCV 2024] EchoScene: Indoor Scene Generation via Information Echo over Scene Graph Diffusion.
60 stars 1 forks source link

Training Time #8

Closed AmingWu closed 2 weeks ago

AmingWu commented 3 months ago

Dear Authors,

How long does your method need to train?

ymxlzgy commented 3 months ago

Hi, Thanks for asking. The full version would be trained for around four or five days on a 40GB GPU. The layout-only branch would be trained for around two days (if I remember correctly). Still depends on the batch size.