search

tobran / GALIP

[CVPR2023] A faster, smaller, and better text-to-image model for large-scale training
MIT License
225 stars 25 forks source link

I only have one GPU #21

Closed sabhiram6 closed 8 months ago

sabhiram6 commented 8 months ago

Hey @tobran I have read your

error

paper and its excellent . I am currently implementing the code . I only have one GPU so in train.sh i changed the nodes to 1 and set CUDA_VISIBLE_DEVICES=0 .

But i am getting " train.py: error: unrecognized arguments: --local-rank=0" .

Could you please tell me what changes i need to make in the code if I have access to only 1 GPU .

Thank You

tobran commented 8 months ago

Thank you for your interest in our work. You can check the pytorch version, if it is more than 2.0, you will encounter this error

sabhiram6 commented 8 months ago

Okay thank you. I used torchrun and now it is working .

VinayHajare commented 7 months ago

Okay thank you. I used torchrun and now it is working .

Can you please help me with training on single GPU, I have tried many times also with torchrun. Please tell what changes did you do with the local_rank

sabhiram6 commented 6 months ago

hey Vinay , I made some small changes like setting MultiGpu to False everywhere and only gave one GPU id in yml files. And later adjusted those according to torchrun .

On Sun, Feb 11, 2024 at 10:22 AM Vinay Arjun Hajare < @.***> wrote:

Okay thank you. I used torchrun and now it is working .

Can you please help me with training on single GPU, I have tried many times also with torchrun. Please tell what changes did you do with the local_rank

— Reply to this email directly, view it on GitHub https://github.com/tobran/GALIP/issues/21#issuecomment-1937424963, or unsubscribe https://github.com/notifications/unsubscribe-auth/AWBIGNEPSBAC2L4FAGB3PW3YTBE7LAVCNFSM6AAAAABBGJHKY2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSMZXGQZDIOJWGM . You are receiving this because you modified the open/close state.Message ID: @.***>