skills.rest
zhuyingqin
zhuyingqinCOMMUNITY
Explore all skills by this author →
GitHub

run-experiment

Automate ML experiment deployment and execution across local, SSH, Vast.ai, and Modal GPUs.

1|1|Updated May 19, 2026
Search Tags:#modal#job orchestration#vast.ai#gpu deployment#ml experimentation#ssh training#wandb logging
One-click install
npx skills add https://github.com/zhuyingqin/ARIS-WEB --skill run-experiment-zhuyingqin
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: run-experiment
Source: https://github.com/zhuyingqin/ARIS-WEB/tree/main/crates/runtime/assets/skills/run-experiment
Command: npx skills add https://github.com/zhuyingqin/ARIS-WEB --skill run-experiment-zhuyingqin

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes assets (resource) components.

What problem does it solve?

It removes the manual, error-prone effort of provisioning GPU resources and launching ML training experiments across local machines, SSH servers, Vast.ai instances, and Modal serverless GPUs.

Core Features & Use Cases

  • Environment-aware deployment: Detects local vs remote (SSH) vs Vast.ai vs Modal based on project CLAUDE.md, then follows the correct launch path.
  • GPU pre-flight checks: Verifies available GPU memory before assignment to avoid waste and immediate runtime failures (where applicable).
  • Remote code sync and job orchestration: Syncs code (rsync or git; Vast.ai always via rsync), installs requirements when present, starts runs via screen, and verifies launch state.
  • Experiment observability hooks: Optionally injects Weights & Biases logging into scripts when enabled in CLAUDE.md.
  • Cost controls for hosted GPUs: Supports Vast.ai auto-destroy after completion and Modal cost-aware execution with timeouts and volumes.

Quick Start

Use the run-experiment skill to start your training job by saying: run experiment and include a description of what model training you want to launch.

Frequently Asked Questions about run-experiment

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

FAQPage Schema
How do I deploy and run GPU training jobs on remote servers?▼

To deploy and run GPU training jobs, this automates execution across local GPUs, SSH remote servers, Vast.ai instances, and Modal serverless GPUs by syncing code, installing requirements, and starting runs via screen sessions.

Does this support Vast.ai auto-destroy and cost controls for rented GPU instances?▼

Yes, Vast.ai auto-destroy is fully supported after training completion. The system also provides Modal cost-aware execution with timeouts and volumes to manage expenses on rented GPU instances.

Can I integrate Weights & Biases logging into my ML experimentation workflow?▼

You can integrate Weights & Biases (wandb) logging into your ML experimentation workflow. The system optionally injects wandb logging hooks into your scripts when enabled in the project CLAUDE.md configuration.

What is the best way to verify GPU availability before launching an ML experiment?▼

The best way to verify GPU availability is through built-in pre-flight checks. The system checks available GPU memory before assignment to prevent resource waste and immediate runtime failures where applicable.

How to sync code and manage job orchestration for SSH training?▼

For SSH training job orchestration, the system syncs your code via rsync or git, automatically installs dependencies when present, starts runs through screen, and verifies the launch state to ensure successful execution.

Do I need a CLAUDE.md file to use environment-aware deployment for ML experiments?▼

Yes, you need a CLAUDE.md file for environment-aware deployment. The system interprets this file to detect local versus remote, Vast.ai, or Modal environments and follows the correct platform-specific launch path.

Related Skills

Explore More Data & Analytics Skills→
HeXiao-55HeXiao-55
1

run-experiment

Automates deployment and execution of ML experiments on local or remote GPU servers.

Community
Advanced
Mang30Mang30

run-experiment

Launches ML training jobs on local, SSH, Vast.ai, or Modal GPUs with logging.

Community
Advanced
goupup-aigoupup-ai
2

run-experiment

Automate ML training experiment deployment across local, remote, Vast.ai, and Modal GPU environments.

Community
Advanced
lightrain-alightrain-a

run-experiment

Deploys ML experiments to local, SSH, Vast.ai, or Modal GPUs with verification.

Community
Advanced