search

onnx / tutorials

Tutorials for creating and using ONNX models
Apache License 2.0
3.39k stars 629 forks source link

Could I build my own model and wrap with ONNX runtime and train under the runtime? #240

Closed laurence-lin closed 3 years ago

laurence-lin commented 3 years ago

The documents note that ONNX provides custom trained model by AutoML service from two Azure cloud.

Now if I have my own model, built under PyTorch or Tensorflow framework, could I wrap in ONNX framework and train under the runtime?

The tutorial notebook demonstrate to execute the inference under these frameworks, and the speed up for making prediction is amazing, but I would like to know if this works for training process.

Thank you!

pranavsharma commented 3 years ago

onnxruntime does have limited training support. Please file an issue under onnxruntime github issues if you need any help.

laurence-lin commented 3 years ago

onnxruntime does have limited training support. Please file an issue under onnxruntime github issues if you need any help.

Thank you, I will look for it.