Openml Tools#

Configuration File: openml_tools.json Tool Type: Local Tools Count: 2

This page contains all tools defined in the openml_tools.json configuration file.

Available Tools#

OpenML_get_dataset (Type: BaseRESTTool)#

Get detailed metadata for a specific OpenML dataset by its numeric ID. Returns comprehensive info…

OpenML_get_dataset tool specification

Tool Information:

  • Name: OpenML_get_dataset

  • Type: BaseRESTTool

  • Description: Get detailed metadata for a specific OpenML dataset by its numeric ID. Returns comprehensive information including description, data format, creator, collection date, license, download URLs (ARFF and Parquet), default target attribute, tags, citation info, and version history. Well-known dataset IDs include: 61 (iris), 40996 (Fashion-MNIST), 554 (mnist_784), 1169 (airlines), 31 (credit-g), 1590 (adult). Use OpenML_search_datasets to discover dataset IDs.

Parameters:

  • data_id (integer) (required) OpenML dataset ID number (e.g., 61 for iris, 554 for MNIST, 31 for credit-g). Obtain from OpenML_search_datasets.

Example Usage:

query = {
    "name": "OpenML_get_dataset",
    "arguments": {
        "data_id": 10
    }
}
result = tu.run(query)

OpenML_search_datasets (Type: BaseRESTTool)#

List/discover OpenML benchmark datasets in catalog order. This endpoint does not perform keyword …

OpenML_search_datasets tool specification

Tool Information:

  • Name: OpenML_search_datasets

  • Type: BaseRESTTool

  • Description: List/discover OpenML benchmark datasets in catalog order. This endpoint does not perform keyword search. Results use path-based offset pagination: start with offset=0 and continue with metadata.next_offset while metadata.has_more is true. OpenML exposes no total here; ToolUniverse reports an exact total only on a short final page. Returns dataset IDs, names, versions, formats, and quality statistics.

Parameters:

  • limit (integer) (optional) Datasets per page (default 20, max 10000).

  • offset (integer) (optional) Zero-based dataset offset. Continue with metadata.next_offset while metadata.has_more is true.

Example Usage:

query = {
    "name": "OpenML_search_datasets",
    "arguments": {
    }
}
result = tu.run(query)