search

Qsingle / LearnablePromptSAM

Try to use the SAM-ViT as the backbone to create the learnable prompt for semantic segmentation
Apache License 2.0
77 stars 13 forks source link

Video memory requirement #1

Open WusterHappy opened 1 year ago

WusterHappy commented 1 year ago

May I ask how much video memory your project needs at least to run? Is 24g 3090 enough

Qsingle commented 1 year ago

Good question, 24 GB memory is enough for all of the models. If you use the vit-b, 12GB memory can run the model.

krishnaadithya commented 1 year ago

Is it enough? I am trying to train the vit-b in 24gb GPU model and it only train in bs 1

Qsingle commented 1 year ago

Yeah, it is enough. For the ViT-H, bs 1 is the largest batch size we can run on one GPU. You can try other backbone like ViT-B or ViT-L