We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
It seems you have implemented multiple gpus for tensorflow backend:
from imageai.Detection.Custom import DetectionModelTrainer trainer = DetectionModelTrainer() trainer.setModelTypeAsYOLOv3() trainer.setDataDirectory(data_directory="/content/gdrive/MyDrive/ai-training/test") trainer.setGpuUsage(1) trainer.setTrainConfig(object_names_array=["Item 1", "Sample 2"], batch_size=4, num_experiments=200) trainer.trainModel()
However, this has not yet been implemented for the pytorch backend. Is this something you intend to add in future?
Thank you!
The text was updated successfully, but these errors were encountered:
Added to to list of to do in next update
Sorry, something went wrong.
No branches or pull requests
It seems you have implemented multiple gpus for tensorflow backend:
from imageai.Detection.Custom import DetectionModelTrainer
trainer = DetectionModelTrainer()
trainer.setModelTypeAsYOLOv3()
trainer.setDataDirectory(data_directory="/content/gdrive/MyDrive/ai-training/test")
trainer.setGpuUsage(1)
trainer.setTrainConfig(object_names_array=["Item 1", "Sample 2"], batch_size=4, num_experiments=200)
trainer.trainModel()
However, this has not yet been implemented for the pytorch backend. Is this something you intend to add in future?
Thank you!
The text was updated successfully, but these errors were encountered: