finetuning

Fine-tune Large Language Models on Azure AI Foundry using SFT, DPO, or RFT.

Updated Aug 27, 2026
One-click install
npx skills add https://github.com/mordechai30/.agents --skill finetuning-mordechai30
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: finetuning
Source: https://github.com/mordechai30/.agents/tree/main/skills/microsoft-foundry/finetuning
Command: npx skills add https://github.com/mordechai30/.agents --skill finetuning-mordechai30

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires openai, azure-identity, azure-ai-projects, requests, pandas, pyarrow, and includes scripts (resource) and references (resource) and workflows (resource) components.

What problem does it solve?

This skill streamlines the complex lifecycle of fine-tuning Large Language Models on Azure AI Foundry, removing the friction from dataset preparation, training, and deployment.

Core Features & Use Cases

  • Multi-Method Training: Supports SFT (supervised), DPO (preference), and RFT (reinforcement with graders) to optimize model behavior.
  • End-to-End Pipeline: Provides automated tools for data validation, training job submission, hyperparameter tuning, and model evaluation.
  • Use Case: Use this skill to train a specialized model on your company's proprietary documentation or to refine a reasoning model for agentic tool-calling tasks.

Quick Start

Use the finetuning skill to submit a supervised fine-tuning job using your training file train.jsonl and validation file val.jsonl.

Frequently Asked Questions about finetuning

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

FAQPage Schema
How do I fine-tune a Large Language Model on Azure AI Foundry?

To fine-tune an LLM on Azure AI Foundry, this skill automates the pipeline from dataset validation to deployment using SFT, DPO, or RFT methodologies. It manages training job submission and hyperparameter tuning.

What is the difference between SFT, DPO, and RFT for model training?

SFT (supervised), DPO (preference), and RFT (reinforcement with graders) are distinct methodologies to optimize model behavior. This skill supports all three for refining reasoning or adapting models to proprietary documentation.

Can I use JSONL files for supervised fine-tuning on Azure?

Yes, you can use JSONL files for supervised fine-tuning on Azure. The skill processes training and validation files, providing automated dataset conversion and validation before submitting the SFT job.

Does this skill handle model deployment after training?

Yes, the skill manages model deployment after training. It handles the entire end-to-end pipeline, deploying the custom model on Azure infrastructure once the SFT, DPO, or RFT training job completes.

What dependencies are required to run Azure AI fine-tuning workflows?

Required dependencies include openai, azure-identity, azure-ai-projects, requests, pandas, and pyarrow. These libraries enable dataset manipulation, Azure authentication, and training job submission.