AGENT SKILL

finetuning

Fine-tune models on Microsoft Foundry using SFT (supervised), DPO (preference), or RFT (reinforcement with graders). Covers dataset preparation, training job submission, deployment, and evaluation. USE FOR: fine-tune, SFT, DPO, RFT, training data, grader, distillation, fine-tuned model, training job, large file upload, calibrate grader, deploy fine-tuned model, evaluate fine-tuned model. DO NOT USE FOR: general model deployment without fine-tuning (use deploy-model), agent creation (use agents), prompt optimization without training (use prompt-optimizer).

License 许可
MIT
最近更新
2026-11-14
Stars
★ 1.6k
Forks
267

01 前言 Front Matter

name
finetuning
license
MIT
metadata
{ "author": "Microsoft", "version": "0.0.0-placeholder" }
description
Fine-tune models on Microsoft Foundry using SFT (supervised), DPO (preference), or RFT (reinforcement with graders). Covers dataset preparation, training job submission, deployment, and evaluation. USE FOR: fine-tune, SFT, DPO, RFT, training data, grader, distillation, fine-tuned model, training job, large file upload, calibrate grader, deploy fine-tuned model, evaluate fine-tuned model. DO NOT USE FOR: general model deployment without fine-tuning (use deploy-model), agent creation (use agents), prompt optimization without training (use prompt-optimizer).

02 资源文件 33

03 SKILL.md 正文

Fine-Tuning on Microsoft Foundry

Fine-tune models using SFT (supervised), DPO (preference), or RFT (reinforcement with graders). Covers dataset prep, training, deployment, and evaluation.

When to Use

Use this sub-skill when the user asks about:

  • Fine-tuning a model (SFT, DPO, or RFT)
  • Preparing, validating, or formatting training data
  • Submitting, monitoring, or diagnosing training jobs
  • Calibrating graders or pass thresholds for RFT
  • Deploying or evaluating a fine-tuned model
  • Choosing between training types (SFT vs DPO vs RFT)
  • Distillation, synthetic data generation, or dataset quality scoring
  • Large file uploads for training data
  • Cleaning up fine-tuning resources (files, deployments)

Do NOT use for: General model deployment without fine-tuning (use deploy-model), agent creation (use agents), prompt optimization without training (use prompt-optimizer).

Workflows

Stage Guide
Quick start workflows/quickstart.md
Full pipeline workflows/full-pipeline.md
Create data workflows/dataset-creation.md
Iterate workflows/iterative-training.md
Diagnose workflows/diagnose-poor-results.md

References

Topic File
SFT vs DPO vs RFT references/training-types.md
Hyperparameters references/hyperparameters.md
Data formats references/dataset-formats.md
Grader design (RFT) references/grader-design.md
Reward hacking references/reward-hacking.md
Agentic RFT (tools) references/agentic-rft.md
Deployment references/deployment.md
Training curves references/training-curves.md
Evaluation references/evaluation.md
Vision fine-tuning references/vision-fine-tuning.md
Large file uploads references/large-file-uploads.md
Platform gotchas references/platform-gotchas.md

Scripts

Script Purpose
scripts/submit_training.py Submit SFT/DPO/RFT jobs
scripts/monitor_training.py Poll job until completion
scripts/calibrate_grader.py Find optimal RFT pass_threshold
scripts/check_training.py Analyze curves, list checkpoints
scripts/deploy_model.py Deploy via ARM REST API
scripts/evaluate_model.py LLM judge evaluation
scripts/convert_dataset.py Convert between SFT/DPO/RFT formats
scripts/generate_distillation_data.py Generate synthetic training data
scripts/score_dataset.py Quality scoring on training data
scripts/cleanup.py Delete old files and deployments
scripts/validate/ Data validators (SFT, DPO, RFT) + stats

Rules

  1. Always baseline first — evaluate the base model before fine-tuning
  2. Validate data before submitting — run scripts/validate/validate_sft.py
  3. Calibrate RFT graders — target 25-50% failure rate on the base model
  4. Evaluate checkpoints — don't blindly deploy the final one
  5. Measure token cost alongside accuracy when comparing models

Quick Reference

Task Command
Validate SFT data python scripts/validate/validate_sft.py data.jsonl
Submit SFT job python scripts/submit_training.py --model gpt-4.1-mini --training-file train.jsonl --validation-file val.jsonl --type sft
Monitor job python scripts/monitor_training.py --job-id ftjob-xxx
Analyze curves python scripts/check_training.py --job-id ftjob-xxx
Deploy model python scripts/deploy_model.py --model-id ft:gpt-4.1-mini:... --name my-eval
Evaluate model python scripts/evaluate_model.py --deployment-name my-eval --test-file test.jsonl

Error Handling

Error Cause Fix
"API version not supported" Older openai SDK on /v1/ endpoint Upgrade to openai>=1.0
"does not support fine-tuning with Standard TrainingType" OSS model needs globalStandard Use --use-rest flag or script auto-falls back
Job stuck in post-training eval Under-provisioned tool endpoint (RFT) Scale to S2+, enable Always On
"DeploymentNotReady" after ARM succeeds ARM/data-plane race condition Delete and recreate deployment, wait 5 min
Content safety block at deployment PII-dense training data Remove problematic document types

04 同仓库更多 Skills

microsoft
microsoft/azure-skills

azure-advisor-cost-optimization

Analyze Azure Advisor Cost/IdleResources findings and prevalidate deletion safety, including Service Groups. WHEN: "find unused Azure resources", "what resources am I paying for but no longer using", "idle-resource savings", "historical idle-resource savings", "are these idle resources safe to delete". Generic Cost, other subcategories, and cross-category requests use azure-advisor.

★ 1.6k
microsoft
microsoft/azure-skills

python-appservice-deploy

Deploy Python (Flask/Django/FastAPI) code to Azure App Service Linux. WHEN: "Flask App Service", "Django App Service", "FastAPI App Service", "deploy Python to App Service". DO NOT USE FOR: Container Apps, Functions, non-Python, Terraform/Bicep/IaC, full infra — use azure-prepare.

运维与部署
★ 1.6k
microsoft
microsoft/azure-skills

preset

Intelligently deploys Azure OpenAI models to optimal regions by analyzing capacity across all available regions. Automatically checks current region first and shows alternatives if needed. USE FOR: quick deployment, optimal region, best region, automatic region selection, fast setup, multi-region capacity check, high availability deployment, deploy to best location. DO NOT USE FOR: custom SKU selection (use customize), specific version selection (use customize), custom capacity configuration (use customize), PTU deployments (use customize).

运维与部署Skill与智能体开发
★ 1.6k
microsoft
microsoft/azure-skills

customize

Interactive guided deployment flow for Azure OpenAI models with full customization control. Step-by-step selection of model version, SKU (GlobalStandard/Standard/ProvisionedManaged), capacity, RAI policy (content filter), and advanced options (dynamic quota, priority processing, spillover). USE FOR: custom deployment, customize model deployment, choose version, select SKU, set capacity, configure content filter, RAI policy, deployment options, detailed deployment, advanced deployment, PTU deployment, provisioned throughput. DO NOT USE FOR: quick deployment to optimal region (use preset).

Skill与智能体开发
★ 1.6k
microsoft
microsoft/azure-skills

capacity

Discovers available Azure OpenAI model capacity across regions and projects. Analyzes quota limits, compares availability, and recommends optimal deployment locations based on capacity requirements. USE FOR: find capacity, check quota, where can I deploy, capacity discovery, best region for capacity, multi-project capacity search, quota analysis, model availability, region comparison, check TPM availability. DO NOT USE FOR: actual deployment (hand off to preset or customize after discovery), quota increase requests (direct user to Azure Portal), listing existing deployments.

★ 1.6k
microsoft
microsoft/azure-skills

deploy-model

Unified Azure OpenAI model deployment skill with intelligent intent-based routing. Handles quick preset deployments, fully customized deployments (version/SKU/capacity/RAI policy), and capacity discovery across regions and projects. USE FOR: deploy model, deploy gpt, create deployment, model deployment, deploy openai model, set up model, provision model, find capacity, check model availability, where can I deploy, best region for model, capacity analysis. DO NOT USE FOR: listing existing deployments (use foundry_models_deployments_list MCP tool), deleting deployments, agent creation (use agent/create), project creation (use project/create).

运维与部署Skill与智能体开发
★ 1.6k