skills.rest
lix965996-art
lix965996-artCOMMUNITY
Explore all skills by this author →
GitHub

run-experiment

Deploy and run machine learning experiments on local or remote GPU hardware.

1|Updated May 14, 2026
Search Tags:#ssh#log management#experiment tracking#ml#wandb#training jobs#gpu deployment
One-click install
npx skills add https://github.com/lix965996-art/MMM --skill run-experiment-lix965996-art
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: run-experiment
Source: https://github.com/lix965996-art/MMM/tree/main/resources/app/skills/run-experiment
Command: npx skills add https://github.com/lix965996-art/MMM --skill run-experiment-lix965996-art

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It removes the friction of launching ML experiments on local or remote GPU machines by automatically handling environment detection, GPU availability checks, code synchronization, deployment, and post-launch verification.

Core Features & Use Cases

  • Environment-aware deployment: Detects local vs remote execution from the project’s CLAUDE.md (or falls back to environment variables / local execution) and prepares the right run commands.
  • GPU pre-flight validation: Checks GPU availability (nvidia-smi for NVIDIA; MPS availability check for Mac) and only proceeds when sufficient free memory is likely available.
  • Remote-ready code syncing: Syncs code to the server via rsync (default) or via git push/pull when configured, avoiding unnecessary large files.
  • Optional experiment instrumentation: When enabled in CLAUDE.md, ensures W&B logging exists in training scripts and verifies wandb login on the target machine.
  • Robust launch workflow: Uses screen sessions for remote jobs and background execution for local long runs, writing logs to files and verifying the launch afterwards.

Quick Start

Ask the skill to run experiment with an appropriate description so it deploys to the configured server (or runs locally) and starts a GPU-bound training process with logs.

Frequently Asked Questions about run-experiment

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

FAQPage Schema
How do I run ML training jobs on a remote GPU server via SSH?▼

To run ML training jobs on a remote GPU server via SSH, the skill detects the execution environment, validates GPU availability, syncs code remotely via rsync or git, creates a screen session, and verifies the launch before reporting the runtime.

How do I deploy ML experiments to GPUs and track metrics with Weights & Biases?▼

You can deploy ML experiments to GPUs and track metrics with Weights & Biases by enabling W&B logging in CLAUDE.md, ensuring the target machine has a verified wandb login, and launching the training process through the skill.

Can I launch local GPU training jobs in the background without blocking my terminal?▼

Yes, you can launch local GPU training jobs in the background without blocking your terminal by using the skill's background execution workflow, which writes logs to files and verifies the launch status for long runs.

Does this workflow check GPU memory availability before starting machine learning training?▼

Yes, this workflow checks GPU memory availability before starting machine learning training by running nvidia-smi for NVIDIA hardware or checking MPS availability for Mac, only proceeding when sufficient free memory is likely available.

What is the best way to sync code to a remote GPU server before running an experiment?▼

The best way to sync code to a remote GPU server before running an experiment is via rsync by default, or through git push and pull when configured, avoiding the transfer of unnecessary large files.

Why does my remote ML training deployment require a screen session?▼

Your remote ML training deployment requires a screen session to ensure the long-running training process remains active and persistent on the remote GPU hardware after you disconnect your SSH connection.

Related Skills

Explore More Data & Analytics Skills→
Mang30Mang30

run-experiment

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

Community
Advanced
thisismynewfmail-uithisismynewfmail-ui

weights-and-biases

Log machine learning experiment metrics and visualize training progress with W&B.

Community
Intermediate
wanshuiyinwanshuiyin
14.4k

run-experiment

Deploy and execute ML experiments on GPU servers via rsync or git with screen sessions.

Community
Advanced
pluginagentmarketplacepluginagentmarketplace
3

data-ai

Automate Linux data science environment setup with GPU drivers and ML frameworks.

Community
Advanced