skills.rest
nebius
nebiusOFFICIAL
Explore all skills by this author →
WebsiteGitHub

cosmos3-inference

Execute Cosmos3 text-to-image inference workflows on Nebius GPU clusters.

17|8|Updated Apr 7, 2026
Search Tags:#gpu#inference#synthetic-data#skypilot#physical-ai#cosmos3
One-click install
npx skills add https://github.com/nebius/nebius-physical-ai --skill cosmos3-inference
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: cosmos3-inference
Source: https://github.com/nebius/nebius-physical-ai/tree/main/skills/workflows/cosmos3-inference
Command: npx skills add https://github.com/nebius/nebius-physical-ai --skill cosmos3-inference

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill streamlines the execution of complex Cosmos3 inference tasks on Nebius infrastructure, ensuring consistent prompt handling, guardrail enforcement, and output management without manual configuration overhead.

Core Features & Use Cases

  • Automated Inference: Runs text-to-image smoke workflows using the Cosmos framework on H100 GPU clusters.
  • Guardrail Management: Provides controlled access to inference guardrails, allowing for secure and compliant model generation.
  • Use Case: A researcher needs to generate a series of synthetic images for a robotics training dataset; this skill automates the environment setup, model inference, and S3 upload process in a single declarative workflow.

Quick Start

Use the cosmos3-inference skill to run a text-to-image generation task with the specified prompt and default guardrail settings.

Frequently Asked Questions about cosmos3-inference

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

FAQPage Schema
How do I run text-to-image inference workflows on GPU clusters using SkyPilot?▼

You can execute text-to-image inference workflows on Nebius H100 GPU clusters using SkyPilot orchestration. This skill automates environment configuration, prompt handling, and S3 output synchronization for synthetic data generation tasks.

How does synthetic data generation work with the NVIDIA Cosmos framework?▼

Synthetic data generation with the Cosmos framework involves running text-to-image inference on GPU clusters. The process enforces guardrail policies for secure model serving and manages outputs to ensure compliant, automated dataset creation.

Can I use Nebius GPU clusters for generating robotics training datasets?▼

Yes, Nebius H100 GPU clusters support generating synthetic images for robotics training datasets. The workflow automates environment setup, model inference, and S3 upload in a single declarative process tailored for physical AI applications.

What's the best way to manage guardrails during text-to-image model serving?▼

Managing guardrails during model serving requires controlled access to inference policies. This skill provides built-in guardrail management, ensuring secure and compliant synthetic image generation on Nebius infrastructure.

Do I need S3 storage to synchronize outputs from Cosmos3 inference workflows?▼

S3 storage is optional for output synchronization in Cosmos3 inference workflows. When enabled, it automatically uploads generated synthetic data, streamlining dataset management after text-to-image tasks complete on GPU clusters.

Related Skills

Explore More Software Engineering Skills→
JuanMS20JuanMS20

tensorrt-llm

Optimize LLM inference on NVIDIA GPUs with TensorRT-LLM.

Community
Advanced
nebiusnebius
17

dataset

Ingest, validate, and curate sensor data into versioned datasets.

Official
Advanced
nebiusnebius
17

submit-workflow

Validate and submit SkyPilot workflows for Nebius Physical AI clusters.

Official
Advanced
nebiusnebius
17

physical-ai-data-factory

Orchestrate physical AI data workflows with annotation, augmentation, and validation on Nebius.

Official
Advanced