Most Polyaxon work happens in loops: change code or configuration, run again, inspect the result. The right loop depends on where your code lives and how much of the run you want to drive from the CLI, a notebook, or Python.

Use local code uploads when you are editing files on your machine and want Polyaxon to package the current folder for each run.

Use git-based runs when the code should come from a repository. This is the better default for shared projects and reproducible runs.

Use inline scripts for small commands, debugging, or generated code that does not need its own repository.

Use notebook iteration when you want to run experiments from a notebook session and keep the resulting runs tracked in Polyaxon.

Use programmatic iteration when your loop lives in Python and you want to create runs through the client.

Use presets when several runs share the same queues, resources, connections, or environment settings.

---