What are the system requirements for this skill?

It requires Podman for container management and is optimized for environments like Overthink OS that use Quadlet service definitions.

How does this skill handle data persistence?

The skill manages a MongoDB sidecar container for metadata and allows you to mount a workspace directory so your datasets are preserved across container restarts.

Does it support hardware acceleration?

Yes, it includes automatic GPU detection and explicit configuration options for NVIDIA, AMD, and Intel GPUs to speed up model inference within FiftyOne.

Can I run multiple FiftyOne instances at once?

Yes, the skill supports instance numbering (e.g., --instance=2) with automated port allocation to prevent networking conflicts between multiple environments.

What is FiftyOne used for in this skill?

FiftyOne is used to visualize, curate, and audit datasets for machine learning and computer vision tasks, providing a web-based interface to explore data.

FiftyOne Dataset Manager

Name: FiftyOne Dataset Manager
Author: atrawog

byatrawog

0•

Ciencia de Datos y ML

Manages FiftyOne dataset visualization and curation environments using Podman Quadlet containers with integrated MongoDB sidecars.

This skill enables Claude to configure, deploy, and manage FiftyOne, a premier open-source tool for building high-quality datasets and computer vision models. It automates the orchestration of a multi-container architecture involving the FiftyOne web UI and a MongoDB sidecar for persistent storage, specifically optimized for Overthink OS. With built-in support for GPU acceleration and workspace mounting, it streamlines the process of visualizing, auditing, and curating machine learning datasets directly within a containerized environment, making it ideal for ML engineers and data scientists.

Características Principales

01Automated GPU acceleration detection for NVIDIA, AMD, and Intel hardware

02Persistent dataset storage with custom workspace directory mounting

03Built-in FiftyOne plugin management and interactive shell access

04Multi-container orchestration of FiftyOne and MongoDB sidecars

05Streamlined lifecycle management including start, stop, and log interleaving

060 GitHub stars

Casos de Uso

01Curating ML data by identifying labeling errors and removing duplicates

02Visualizing and auditing large-scale computer vision datasets for quality control

03Deploying GPU-accelerated environments for model performance and dataset analysis

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add atrawog/overthink-plugins fiftyone

For use in Claude.ai and ChatGPT

Download Skill