raga-testing-platform

Raga Testing Platform

The raga-testing-platform package provides a Python client for interacting with the Raga Testing Platform. It allows you to easily create and manage test sessions, datasets, and perform various testing operations.

Installation

You can install raga-testing-platform using pip:

pip install raga-testing-platform

Usage

To use the package, import the necessary classes and modules:

from raga import Dataset, TestSession, Auth
import pandas as pd
from typing import Optional, List, Dict

Creating a test DataFrame

test_df = pd.DataFrame({
    'column1': [1, 2, 3],
    'column2': ['a', 'b', 'c']
})

Defining the Schema class

class Schema:
    def __init__(
        self,
        prediction_id: Optional[str] = None,
        timestamp_column_name: Optional[str] = None,
        feature_column_names: Optional[List[str]] = None,
        metadata_column_names: Optional[List[str]] = None,
        label_column_names: Optional[Dict[str, str]] = None,
        embedding_column_names: Optional[Dict[str, str]] = None,
    ):
        self.prediction_id = prediction_id
        self.timestamp_column_name = timestamp_column_name
        self.feature_column_names = feature_column_names
        self.metadata_column_names = metadata_column_names
        self.label_column_names = label_column_names
        self.embedding_column_names = embedding_column_names

Creating an instance of the Schema class

schema = Schema()

Creating an instance of the Auth class

It will create auth token for further usages.

auth = Auth()

This variable stores auth token. You can use it untill expire token.

auth.token

Creating an instance of the TestSession class

In this instance experiment will create and return experiment_id

experiment_id = TestSession(token, 1, "my_experiment")

This variable stores experiment_id. You can use it in further experiment.

experiment_id.experiment_id

Creating an instance of the Dataset class

test_ds = Dataset(token, experiment_id, test_df, schema, "DatasetName")

Loading labels from a file

test_ds.load_labels_from_file(
    "/path/to/labels.json",
    "dataset_name",
    "id_column_name",
    "label_column_name",
    "metadata_column_name",
    "category_column_name",
    "category_id_column_name"
)

To Debug

export DEBUG=1