News for hackers. Stuff that matters.

A comment on Cortex: Deploy machine learning models in production. Parent comment

Re:Cortex: Deploy machine learning models in production by ospillinger on Friday October 18, 2019 @02:36AM (#21289154)
Yes, Cortex uses ONNX Runtime (https://github.com/microsoft/onnxruntime) under the hood so any model that can be exported to ONNX can be deployed.
[ Parent | on HN ]
Re:Cortex: Deploy machine learning models in production by solidasparagus on Friday October 18, 2019 @05:04AM (#21289776)
Is it only able to handle ONNX models? That's a pretty massive limitation compared to a hosted SageMaker endpoint.
[ Parent | on HN ]
Re:Cortex: Deploy machine learning models in production by vishalbollu on Friday October 18, 2019 @08:38AM (#21290847)
Contributor here - Cortex supports Tensorflow saved models in addition to ONNX. PyTorch support is on the roadmap. Do you have specific frameworks in mind that you would like Cortex to support?
[ Parent | on HN ]
Re:Cortex: Deploy machine learning models in production by solidasparagus on Friday October 18, 2019 @09:15PM (#21296232)
Perfect. Nothing in particular other than TF.
[ Parent | on HN ]