Accelerate training via training on multiple gpus

showlab / DatasetDM

[NeurIPS2023] DatasetDM:Synthesizing Data with Perception Annotations Using Diffusion Models

https://weijiawu.github.io/DatasetDM_page/

290 stars 12 forks source link

Accelerate training via training on multiple gpus #22

Open Championchess opened 6 months ago

Championchess commented 6 months ago

Thanks for your great work! May I wonder whether there is a way to train the model parallelly on multiple gpus so that the training could be speeded up?

weijiawu commented 4 months ago

Sorry, we haven't optimized this part as we only trained on a small amount of data.