Skip to content

A Question About Training Time #13

@metphi

Description

@metphi

Thanks for your contribution!

I want to train a new model in an another dataset. But I don't how much the gpu and the time will cost.

I tried to reproduce the training step in DIV2K so I use the code "python train.py --config configs/train_edsr-sronet.yaml --gpu 0,1,2". And I found that it would cost 70h in 3 NVDIA RTX 4090. But in your work that mentioned "All tests were conducted using a single NVIDIA RTX 3090.". I want to know if it is normal that training a edsr-baseline-SRNO model in DIV2K will cost about 70h in three 4090.

Thanks!

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions