Training with a smaller training set

bknyaz / sgg

Train Scene Graph Generation for Visual Genome and GQA in PyTorch >= 1.2 with improved zero and few-shot generalization.

https://arxiv.org/abs/2007.05756

Other

130 stars 20 forks source link

Open ranran9991 opened 3 years ago

ranran9991 commented 3 years ago

Is there any way to decrease the size of the training set? (specifically when using GQA)