Skip to main content

Training

trolo provides flexible training options through both CLI and Python API interfaces.

CLI Usage​

Basic training command:

trolo train --config dfine_n

Multi-GPU Training​

trolo automatically handles distributed data parallel (DDP) training:

trolo train --config dfine_n --device 0,1,2,3

Python API​

from trolo.trainers import DetectionTrainer

# Initialize trainer
trainer = DetectionTrainer(config="dfine_n")

# Start training
trainer.train()

# Multi-GPU training
trainer.train(device="0,1,2,3")

Configuration​

Training configurations can be customized through YAML files. See the configuration for details.