Skip to content
Get packs

Agent Platform Model Tuning

Guides an agent through fine-tuning open or Gemini models on Google Cloud's tuning service, step by step.

Source Download skill
Repository
google/skills
Stars
21,782
Licence
Apache-2.0
Files
12 files

What it does

Once installed, the agent follows a structured workflow for customizing an AI model on Google Cloud: checking your project and region, confirming whether you want an open model or a Gemini model, preparing and validating your training data (JSONL, uploaded to a Cloud Storage bucket you already own), estimating cost, submitting a tuning job, monitoring it, and optionally deploying the result. It includes reference catalogs of supported models and helper Python scripts for listing models, estimating cost, preparing datasets, submitting, monitoring, and cancelling jobs. It is designed to pause and ask you before taking actions rather than acting on its own.

Great for

  • Developers who already have a Google Cloud project and want guided help fine-tuning an open or Gemini model
  • Data teams preparing training datasets in the JSONL format the tuning service expects
  • Teams who want a cost estimate before launching a paid tuning job

Skip it if

  • People without a Google Cloud account or billing set up, since tuning jobs cost money and require cloud permissions
  • Anyone wanting to train a model from scratch or outside Google's platform (the skill explicitly excludes this)
  • Deploying or managing serving endpoints, which the skill says belongs to other separate skills

Try asking

  • “Help me fine-tune Gemma 3 on my customer support transcripts in project my-project, region us-central1”
  • “Check whether my Google Cloud project is ready for model tuning (APIs, permissions, service agents)”
  • “Estimate the cost of tuning Gemini 2.5 Flash on my 5,000-example JSONL dataset before I run it”
  • “Monitor my running tuning job and cancel it if something goes wrong”
Step-by-step guide: how to use Agent Platform Model TuningInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 12 original files, the Apache-2.0 licence and an attribution note.

    agent-platform-tuning.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip agent-platform-tuning.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip agent-platform-tuning.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Provide your project ID, region, and whether you want an open or Gemini model up front, so the agent does not stop to ask.
  • Review the column mapping and sample rows the agent shows you before any upload or job submission.
  • Ask for the cost estimate first and confirm the bucket and project are the ones you intend to bill.

What's inside

  • references/data_prep.md
  • references/hf_datasets.md
  • references/models.md
  • references/requirements.txt
  • references/tuning_guide.md
  • scripts/calculate_cost.py
  • scripts/cancel_tuning_job.py
  • scripts/list_models.py
  • scripts/monitor_tuning_job.py
  • scripts/prepare_dataset.py
  • scripts/tune_open_model.py
  • SKILL.md

Redistributed unmodified from google/skills under the Apache-2.0 licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.