I am trying to find a way to run python code outside of the Snowflake UI (VS Code specifically) that runs on the pytorch container services. If we use a SF notebook, there is an option to select the container service and then a compute pool of our choice (see screenshot). Is there a way external to the notebook UI to run python code against this same container GPU pool?
It's not currently possible to do this with the Generally Available release of ML Runtime and Notebook. However Snowflake does have this functionality available in Private Preview utilising a 'Headless' mode of execution, where you can initiate execution from an external client e.g. VSCode. If your interested in participating in the Private Preview you can reach out to your Snowflake Sales/Technical representative who can assist in getting you added to the Private Preview program.