Skip to main content

Python API Reference

The repx-py library provides a high-level interface for querying experiment results and loading artifacts.

repx_py.Experiment​

The entry point for interacting with a RepX Lab.

from repx_py import Experiment
exp = Experiment(lab_path="./result")

Constructor​

  • Experiment(lab_path: str | Path)
    • Initialize with the path to the built Lab directory (usually ./result).

Properties​

  • runs (Dict[str, RunView]): Access runs by name.
  • jobs (JobCollection): A collection of all jobs across all runs.

repx_py.RunView​

Represents a single named run (e.g., "simulation").

Properties​

  • name (str): The run name.
  • jobs (JobCollection): A collection of jobs belonging to this run.

repx_py.JobCollection​

A queryable collection of jobs. It behaves like a list but provides filtering capabilities.

Methods​

  • filter(**kwargs) -> JobCollection
    • Filter jobs by parameter values or metadata.
    • Example: exp.jobs.filter(model="resnet", learning_rate=0.01)
  • to_dataframe() -> pandas.DataFrame
    • Returns a DataFrame containing metadata (status, parameters, paths) for all jobs in the collection.
  • __iter__()
    • Iterate over JobView objects.
  • __len__()
    • Count of jobs.

repx_py.JobView​

Represents a single job execution (a concrete instance of a Stage).

Properties​

  • id (str): The unique hash ID of the job.
  • name (str): The stage name.
  • pname (str): The full unique name (including scatter indices).
  • status (str): Current status (COMPLETED, FAILED, etc.).
  • effective_params (dict): The resolved parameters used for this job.

Data Access Methods​

  • get_output_path(output_name: str) -> Path
    • Get the absolute path to a specific output file.
  • load_csv(output_name: str, **kwargs) -> pandas.DataFrame
    • Load an output CSV file directly into Pandas. kwargs are passed to pd.read_csv.
  • load_json(output_name: str) -> dict
    • Load an output JSON file.
  • load_text(output_name: str) -> str
    • Read a text file.