search

modelscope / DiffSynth-Studio

Enjoy the magic of Diffusion models!
Apache License 2.0
6.32k stars 564 forks source link

Multi-GPU support via Accelerate/Deep Speed? #119

Closed lschaupp closed 1 month ago

lschaupp commented 1 month ago

See the headline -> for training LORA/Finetune

Artiprocher commented 1 month ago

Multi-GPU is supported, you can enable it by setting CUDA_VISIBLE_DEVICES="0,1,2,3,4,5,6,7". The training script will automatically leverage all available GPUs for training.