Giters

uta-smile / TCL

code for TCL: Vision-Language Pre-Training with Triple Contrastive Learning, CVPR 2022

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

About GPU Usage and Training Time

yangbang18 opened this issue · comments

commented

Hi, thanks for your great work and code sharing.

According to config/Pretrain.yaml, the batch_size is set to 64 (i.e., each GPU will process 64 image-text pairs) during pretraining. I would like to know how much GPU memory will be used and how much time will be taken per epoch (on 4M dataset) under this setting.

By the way, I have read your excellent paper but can not find the supplementary materials on the web. Could you share a link to download it? Thanks a lot.

commented

Hi, thanks for your interest in our paper.

  1. The GPU memory usage is about 30GB on each GPU. We use A100 which contains 8 GPUs.
  2. The training time for each epoch is about 2 hours.
  3. We don't have the supplementary file. If you need anything that is not clearly mentioned in the paper, just let me know.

Thanks.