load-dataset

Download and convert the Halluminate/WebBench CSV dataset to JSONL with filters.

4|1|Updated Feb 3, 2026

Search Tags:#benchmark #csv #filters #jsonl #dataset #webbench

Ready to run

npx skills add https://github.com/lespaceman/athena-workflow-marketplace --skill load-dataset

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: load-dataset
Source: https://github.com/lespaceman/athena-workflow-marketplace/tree/main/plugins/web-bench/skills/load-dataset
Command: npx skills add https://github.com/lespaceman/athena-workflow-marketplace --skill load-dataset

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Download and prepare the Halluminate/WebBench dataset for benchmarking. This skill automates the process of downloading the CSV, converting it to JSONL, and applying optional filters so you can target specific categories, sample sizes, or website allowlists/blocklists. It uses curl and Node.js with zero Python dependencies, and notes that it does NOT execute tasks—you should use execute-task for actual task execution.

Core Features & Use Cases

CSV to JSONL conversion: Download the Halluminate/WebBench CSV and convert it into a JSONL task list suitable for benchmarking.
Targeted filtering: Apply category, sample size, and website allowlist/blocklist filters to tailor benchmarks.
Output artifact: Writes web-bench-tasks.jsonl to the working directory for immediate downstream use.
Zero-Python dependency: Requires only curl and Node.js, avoiding Python-based tooling.

Quick Start

Download the WebBench CSV and convert it to web-bench-tasks.jsonl in the working directory.

Frequently Asked Questions about load-dataset

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I convert a HuggingFace CSV dataset to JSONL for benchmarking?▼

You can convert a HuggingFace CSV dataset to JSONL for benchmarking by downloading the Halluminate/WebBench CSV and transforming it into a web-bench-tasks.jsonl task list using Node.js and curl.

Can I filter the WebBench dataset by category and sample size?▼

You can filter the WebBench dataset by applying optional category, sample size, and website allowlist or blocklist parameters to tailor the generated JSONL benchmark tasks to specific scenarios.

Do I need Python to prepare WebBench benchmark data?▼

You do not need Python to prepare WebBench benchmark data. The dataset preparation requires only curl and Node.js, avoiding any Python-based tooling dependencies entirely.

Does this tool execute the downloaded benchmark tasks?▼

This tool does not execute benchmark tasks. It strictly downloads and transforms the WebBench CSV into a JSONL task list; you should use a separate execute-task process for actual task execution.

What is the best way to target specific websites in a WebBench benchmark?▼

The best way to target specific websites in a WebBench benchmark is to apply website allowlists and blocklists during the CSV to JSONL conversion, filtering the output JSONL file accordingly.

Where does the prepared WebBench JSONL output file get saved?▼

The prepared WebBench JSONL output file is written directly to your current working directory as web-bench-tasks.jsonl, making it immediately available for downstream benchmark execution.