Hi, i am working on creating retraining pipelines in production. The way i'm doing this is to install clearml-server on my production. Then i recreate the ingestion, preprocessing and training/opt tasks into a clearml-pipeline. Thereafter, i would call clearml-task pipelinename command to trigger the retraining.
I have a few issues with this approach though.
Codes are not allowed in production due to our policies, how might i create clearml-tasks without the python codes exposed? Is there a way to 'export' clearml-tasks from my devt clearml-server and import into prod clearml-server? Or is there a best practice to create this retraining pipeline?
Hi CostlyOstrich36 , thanks. I will check with the Enterprise team then.
So does it mean you basically recreated your entire training environment just in production?
Regarding the questions:
I'm not sure I understand. If you don't expose code... What would be executed? I think this is something available only on the Scale/Enterprise level
