ferries

Manage and monitor automated integration test runs for the Marin framework.

1.2k|146|Updated Mar 22, 2024
One-click install
npx skills add https://github.com/marin-community/marin --skill ferries
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ferries
Source: https://github.com/marin-community/marin/tree/main/.agents/skills/ferries
Command: npx skills add https://github.com/marin-community/marin --skill ferries

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill automates the execution and monitoring of critical integration test runs for the Marin framework, ensuring the stability and reliability of its core functionalities.

Core Features & Use Cases

  • Automated Ferry Runs: Manages both fast, low-cost 'canary' health checks and more comprehensive 'daily' integration runs.
  • Regression Detection: Catches infrastructure and pretraining regressions early.
  • Reproducibility: Ensures consistent data assumptions and monitoring across runs.
  • Use Case: Launch and monitor the daily integration test for the Marin framework, ensuring that recent code changes have not introduced regressions before they impact production.

Quick Start

Use the ferries skill to launch the daily integration run.

Frequently Asked Questions about ferries

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

FAQPage Schema
How do I run automated integration tests for the Marin framework?

Regression testing for Marin is managed by executing canary and daily integration test lanes. Canary runs provide fast, low-cost health checks, while daily runs offer comprehensive testing to catch infrastructure and pretraining regressions early.

How do I monitor canary runs to catch regressions in Marin?

Canary runs for Marin are fast, low-cost health checks that detect early regressions. Daily runs are comprehensive integration tests ensuring broader stability, reproducibility, and consistent data assumptions across the framework.

Do I need specific configurations for accelerators and cluster environments to run Marin integration tests?

Yes, specific configurations for accelerators and cluster environments are required. The ferries skill needs these configurations to properly execute integration test scripts and monitor the Marin framework across the designated infrastructure.

What is the best way to detect pretraining regressions in the Marin framework?

The best way to detect pretraining regressions is by running automated daily integration tests via the ferries skill. This approach ensures reproducibility, validates infrastructure stability, and catches regressions before production impact.

Why does my Marin ferry lane integration test run fail to ensure reproducibility?

Reproducibility in Marin ferry lane runs depends on consistent data assumptions and specific configurations for accelerators and cluster environments. Incorrect setup or changing data assumptions can cause integration test runs to fail or produce inconsistent monitoring results.