This might be a little off-topic here. Assuming that i build a model to my satisfaction using the Clara framework. I have gone through the documentation and see the scripts for train, infer and also the export for generating optimised models. Now, i would like to hand this over to the deployment team. Where i am getting confused is that model is one part of the deployment. should i be writing code and passing that along with the model file to the deployment team? Reason i am asking is that there are thing like sliding window inference. Might be I am missing something here. I have also taken a look at deploy but it seems a lot more obviously. unfortunately, i could not find a document where the model trained from Clara Train is used in deployment. I can at least point out this document to the deployment team where they can take out model and the necessary files and deploy.
Sorry if i asked at wrong place!
Great question! We have full support to easily move trained models in Clara train into Clara deploy. This is how deploy engineers have created most of the pipelines. Clara deploy has app base inference operator which is used in multiple reference pipeline You can check liver ,spleen , covid pipelines.
For steps to move models from clara train follow section 9.5.8. Creating Model Specific Application at app base inference operator
Thank you. I shall try it out.