I wrote a proof-of-concept implementation of AWS Sagemaker multimodel server in Clojure https://github.com/jcpsantiago/sagemaker-multimodel-clj At the moment it supports XGBoost models as I meant it as a proof of concept, and XGB is what I use the most. The included example models were trained in R, because I wanted to use cross-platform compatibility. I’ll try to integrate it with a broader framework to support more models — in Clj land that’s http://tech.ml these days, right? If you use it let me know 🙂 code can be vastly improved for sure, so if you want feel free to drop some PRs
Hi Santiago, I’ve been working on https://github.com/zero-one-group/geni a dataset library that uses Spark as its engine. It has XGBoost and all the other models supported by Spark ML!
@anthony550 that sounds great! I’ll investigate how it could be integrated into this
do you use or have used Sagemaker before actually? I’m having a hard time finding folks using it, which is interesting considering I think it’s a nice value
No, I’m afraid. Will check it out!