experiment-bridge

Implement ML experiments from EXPERIMENT_PLAN.md and deploy to GPUs.

Updated Mar 17, 2026

Search Tags:#automation #workflow #research #orchestration #experiment #ml

Ready to run

npx skills add https://github.com/loujc/Auto-claude-code-research-in-sleep-manual --skill experiment-bridge-loujc

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: experiment-bridge
Source: https://github.com/loujc/Auto-claude-code-research-in-sleep-manual/tree/main/skills/experiment-bridge
Command: npx skills add https://github.com/loujc/Auto-claude-code-research-in-sleep-manual --skill experiment-bridge-loujc

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Bridge ideas to running experiments by turning plans into executable code and initial results, enabling autonomous ML research workflows from concept to deployment.

Core Features & Use Cases

Reads EXPERIMENT_PLAN.md and converts it into runnable experiment code
Automatically deploys experiments to GPUs with sanity checks and progress tracking
Collects and formats initial results for the auto-review loop to critique and iterate
Supports multiple experiment variants and model configurations for ablations

Quick Start

Provide an EXPERIMENT_PLAN.md and run the bridge to automatically implement and deploy experiments.

Frequently Asked Questions about experiment-bridge

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

FAQPage Schema

How do I automate an ML experiment workflow from plan to deployment?▼

You can automate ML experiment workflows by providing an EXPERIMENT_PLAN.md, which the system reads to implement experiment code, deploy to GPUs, and collect initial results for auto-review loops.

How does automatic experiment orchestration handle reproducibility?▼

Experiment orchestration enforces reproducibility by applying fixed seeds, using argparse-driven hyperparameters, and formatting all run outputs as structured JSON and CSV files.

Can I run multiple model configurations for ablations automatically?▼

Yes, automated experiment deployment supports multiple experiment variants and model configurations, enabling parallel deployment controls to run and compare ablations efficiently.

What format do I need to start automated ML experiment deployment?▼

You need to provide an EXPERIMENT_PLAN.md file as input, which serves as the blueprint to automatically generate runnable experiment code and deploy it to GPUs.

How do deployed experiments feed results back into an auto-review loop?▼

Deployed experiments automatically collect and format initial results, passing these structured outputs directly into the auto-review loop to critique, iterate, and refine the research workflow.