Does YOLO-Master support multi-GPU (DDP) training?

[leeyg-gnu]

Hi,

I just wanted to ask whether YOLO-Master currently supports multi-GPU (DDP) training, or if it is designed to run on a single GPU only.

Thanks!

[isLinXu]

Hi @leeyg-gnu,

Yes, YOLO-Master currently supports multi-GPU (Distributed Data Parallel, DDP) training. Since the project is compatible with the Ultralytics framework, you can easily enable DDP by specifying multiple GPU indices in the device parameter.

Usage Example
You can refer to the following code snippet to start multi-GPU training:
from ultralytics import YOLO

Load a model

Build a new model from YAML

model = YOLO('cfg/models/master/v0/det/yolo-master-n.yaml')

# Train the model
results = model.train(
    data='coco.yaml',
    epochs=600, 
    batch=256, 
    imgsz=640,
    device="0,1,2,3", # Use multiple GPUs (e.g., 4 GPUs) 
    scale=0.5, 
    mosaic=1.0,
    mixup=0.0, 
    copy_paste=0.1
)

Key Points

• Device Parameter: Simply pass a string like "0,1,2,3" to the device argument to trigger DDP mode automatically.
• Batch Size: Note that the batch size you specify (e.g., 256) is the total batch size across all GPUs.
  Hope this clarifies your question! Let us know if you encounter any issues during setup.