Closed alinapark closed 1 year ago
Hello @alinapark , For the challenge, is it allowd to change the format of model mentioned. from pytorch to tensorflow / onnx ?
@khushpatel2002 yes, feel free to use either
@alinapark Thank you for your reply.
@alinapark, hello! Is this challenge still relevant?
Overview
Create a service that deploys five NLP models for inference, then receives messages through an exposed POST API endpoint, and finally returns inference results (of all five models) in a single response body. Expected deliverable is a service packed in the Docker image.
Your service could be a well-configured framework or a self-made API server; use any ML model deployment tool you see fit. There's no language limitation. The most important here is the reusability of a final project.
All submissions will be deployed on a
g4dn.2xlarge
instance, and the assessment will be performance (mainly, throughput)-basedMore details in the challenge repository
Additional Resources