new_experiment

Scaffold a new experiment dir under experiments/\<yyyymmdd>_\<slug>/.

Emits two files:

config.yaml - a minimal, runnable ExperimentConfig skeleton with placeholders for the model, data, algorithm, and a metadata block where hypothesis / success_metric / benchmark live.
run.py - the 3-line declarative entrypoint researchers run:

from evsys_sdk import Experiment import src # registers project verifiers / metrics / transforms

Experiment.from_yaml("config.yaml").run()

Use via the CLI: evsys new-experiment \<slug>. Programmatically: new_experiment(project_root, slug).

attribute__all__

= ['new_experiment']

funcnew_experiment(project_root, slug, *, today=None) -> Path

Create \<project_root>/experiments/\<yyyymmdd>_\<slug>/ and write config.yaml + run.py inside. Returns the experiment dir.

Refuses if the dir already exists - researchers can pick a different slug or delete the prior one.

paramproject_rootstr | Path

paramslugstr

paramtodaydate | None

= None

Returns

pathlib.Path

func_normalize_slug(slug) -> str

Sanitize into [a-z0-9_-]+.

paramslugstr

Returns

str

func_config_yaml(slug) -> str

paramslugstr

Returns

str

func_run_py() -> str

Returns

str