`nemo_automodel.components.launcher.skypilot.launcher`#

Module Contents#

Launch a recipe job on a cloud VM via SkyPilot.

`_parse_gpus_per_node`	Extract GPU count from an accelerator string like `'A100:8'`.
`_recipe_module_path`

nemo_automodel.components.launcher.skypilot.launcher.logger#: ‘getLogger(…)’

nemo_automodel.components.launcher.skypilot.launcher._parse_gpus_per_node(accelerators: str) → int[source]#

Extract GPU count from an accelerator string like 'A100:8'.

Returns 1 when the string cannot be parsed.

nemo_automodel.components.launcher.skypilot.launcher._recipe_module_path(recipe_target: str, repo_root: str) → str[source]#

class nemo_automodel.components.launcher.skypilot.launcher.SkyPilotLauncher[source]#

Launch a recipe job on a cloud VM via SkyPilot.

_build_command( recipe_target: str, job_conf_path: str, gpus_per_node: int, num_nodes: int, extra_args: Optional[List[str]] = None, ) → str[source]#

launch( config: Dict[str, Any], config_path: pathlib.Path, recipe_target: str, launcher_config: Dict[str, Any], extra_args: Optional[List[str]] = None, ) → int[source]#