view-results

View and analyze Hawk evaluation results via the hawk CLI.

1|Updated May 12, 2018

Search Tags:#evaluation #data-analysis #transcripts #results #hawk #eval-sets

Ready to run

npx skills add https://github.com/tbroadley/dotfiles --skill view-results

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: view-results
Source: https://github.com/tbroadley/dotfiles/tree/main/claude/skills/hawk-view-results
Command: npx skills add https://github.com/tbroadley/dotfiles --skill view-results

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill helps teams quickly access, understand, and analyze Hawk evaluation outputs after a run, including eval sets, evaluations, samples, and transcripts.

Core Features & Use Cases

List Eval Sets: hawk list eval-sets to view available eval sets with IDs, dates, and creators.
List Evaluations and Samples: hawk list evals [EVAL_SET_ID] and hawk list samples [EVAL_SET_ID] to inspect tasks, models, statuses, and sample counts.
Retrieve Transcripts: hawk transcript <UUID> to obtain full conversations, with optional --raw for JSON.
Bulk Access: hawk transcripts <EVAL_SET_ID> to export all transcripts to a directory, with limit controls.

Quick Start

Use the hawk command to list available eval sets, then list evaluations or samples for an eval set, and fetch a transcript for a selected sample.

Frequently Asked Questions about view-results

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

FAQPage Schema

How do I view Hawk evaluation results and scores after a run?▼

To view Hawk evaluation results, use the hawk CLI to list eval sets, evaluations, and samples. You can retrieve status, scores, and metadata for completed runs using list commands.

What is the best way to export all transcripts from an eval set?▼

The best way to export all transcripts from an eval set is using the 'hawk transcripts <EVAL_SET_ID>' command, which downloads full conversations to a directory with optional limit controls.

Can I retrieve raw JSON for a specific evaluation transcript?▼

Yes, you can retrieve raw JSON for a specific evaluation transcript by running the 'hawk transcript <UUID>' command with the optional --raw flag to get the unformatted conversation data.

Do I need the hawk CLI to inspect evaluation samples and metadata?▼

Yes, you need the hawk CLI installed to inspect evaluation samples and metadata. This skill relies on standard hawk command-line outputs to present lists, statuses, and sample counts.

How do I list available Hawk eval sets to find the right ID?▼

To list available Hawk eval sets and find their IDs, run the 'hawk list eval-sets' command, which displays available eval sets along with their creation dates and creators.