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

sim2real-engine

Orchestrate 14-stage VLM-to-RL Sim2Real pipelines on Kubernetes and S3.

17|8|Updated Apr 7, 2026
Search Tags:#kubernetes#orchestration#reinforcement learning#robotics#vlm#sim2real
One-click install
npx skills add https://github.com/nebius/nebius-physical-ai --skill sim2real-engine
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: sim2real-engine
Source: https://github.com/nebius/nebius-physical-ai/tree/main/skills/workbench/sim2real-engine
Command: npx skills add https://github.com/nebius/nebius-physical-ai --skill sim2real-engine

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires ruff, pytest.

What problem does it solve?

This skill resolves the complexity of managing multi-stage Sim2Real loops, ensuring consistent execution across preamble, inner, outer, and finalize phases in a Kubernetes-native environment.

Core Features & Use Cases

  • Canonical Stage Management: Provides a standardized 14-stage map for VLM-to-RL workflows, ensuring alignment between monitor status and S3 artifact contracts.
  • Workflow Orchestration: Automates the transition between preamble, inner loops, and finalization, including K8s sibling job management.
  • Use Case: Use this skill when debugging or modifying the Sim2Real pipeline to ensure that stage-specific artifacts like env generation, policy rollouts, and VLM evaluation are correctly tracked and executed.

Quick Start

Use the sim2real-engine skill to validate the current workflow configuration and check the status of the latest K8s sibling jobs.

Frequently Asked Questions about sim2real-engine

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

FAQPage Schema
How do I orchestrate multi-stage VLM-to-RL sim2real pipelines on Kubernetes?▼

Sim2real pipelines on Kubernetes are orchestrated by managing a 14-stage map across preamble, inner, outer, and finalize phases. This ensures consistent workflow execution, automates K8s sibling job transitions, and maintains artifact integrity on S3 infrastructure.

What is the standard stage map for VLM-to-RL sim2real workflows?▼

The standard stage map for VLM-to-RL sim2real workflows defines 14 specific execution phases. It aligns monitor status reporting with S3 artifact contracts for stage-specific outputs like environment generation, policy rollouts, and VLM evaluation.

Do I need the npa CLI to manage sim2real pipeline orchestration?▼

Yes, managing sim2real pipeline orchestration requires integration with the npa CLI. Adherence to the defined stage map via the CLI ensures consistent workflow deployment, correct artifact tracking, and proper K8s sibling job monitoring.

How do I track S3 artifacts during a reinforcement learning sim2real loop?▼

S3 artifacts during a reinforcement learning sim2real loop are tracked through canonical stage management. The 14-stage map aligns monitor status with artifact contracts, ensuring outputs like policy rollouts and VLM evaluations are correctly verified.

Why does my Kubernetes sim2real pipeline fail during stage transitions?▼

Kubernetes sim2real pipelines fail during stage transitions if the 14-stage map configuration is misaligned. Validating the workflow configuration and checking K8s sibling job status ensures preamble, inner, outer, and finalize phases execute properly.

Related Skills

Explore More Software Engineering Skills→
nebiusnebius
17

sim2real-operate

Manage Sim2Real VLM-to-RL pipeline jobs on Kubernetes GPU clusters.

Official
Advanced
notquenotque
415

article-evaluation-pipeline

Evaluate articles for voice authenticity through a deterministic four-phase pipeline.

Community
Advanced
davidahmanndavidahmann
12

gait-policy-test-rollout

Test Gait policy changes against intent fixtures and simulate staged rollouts.

Community
Intermediate
nebiusnebius
17

foxglove

Integrate the Foxglove embedded viewer and MCAP processing tools into NPA agent workflows.

Official
Advanced