For AI agents: a documentation index is available at the root level at /llms.txt. Append /llms.txt to any URL for a page-level index, or .md for the markdown version of any page.
LogoLogoNeMo Gym
DocumentationAPI Reference
DocumentationAPI Reference
  • About
    • Concepts
    • Architecture
    • Ecosystem
    • Release Notes
  • Get Started
    • Prerequisites
    • Installation
    • Quickstart
  • Prepare Data
    • Prepare and Validate
    • Download from Hugging Face
    • Prompt Config
  • Configure Agents
    • Integrate Existing Agents
    • Drive a Remote Agent
    • Agent Skills
  • Configure Models
    • OpenAI
    • Azure OpenAI
    • Inference Providers
    • Model-call capture
    • vLLM
    • Local vLLM
    • Local vLLM Proxy
    • Anthropic Messages
    • Switchyard
  • Build Verifiers
    • Verification Patterns
    • Multi-Reward Verification
  • Evaluate
    • Benchmarks
    • Compare Harnesses
    • Browse Environments
    • Aggregate Metrics
    • Rollout Health Checks
    • Diagnose Results
  • Tutorials
    • Training Tutorials
      • NeMo RL
      • Unsloth
      • Multi-Environment Training
      • Training with VeRL
      • Offline Training (SFT/DPO)
      • Training on External Agent Harnesses
    • Evaluation Tutorials
  • Build Environments
    • Resources Server APIs
    • Single-Step Environment
    • Multi-Step Environment
    • Stateful Environment
    • MCP Resources Server
    • Real-World Environment
    • Integrate external libraries
  • Model Recipes
    • Nemotron 3 Nano
    • Nemotron 3 Super
  • Infrastructure
    • Deployment Topology
    • Sandbox API
    • Engineering Notes
  • Observability
    • Configuration
    • Distributed Tracing
    • Span Groups
    • Metrics
    • Extending
  • Reference
    • Configuration
    • RL Framework Compatibility
    • CLI Commands
    • FAQ
    • Trajectory capability matrix
  • Troubleshooting
    • Configuration Errors
  • Contribute
    • Development Setup
    • Environments
    • Integrate RL Frameworks
    • Agent Skills
  • About
  • Concepts
  • Environments
  • Evaluation
  • Training
  • Key Terminology
  • Architecture
  • Ecosystem
  • Release Notes
  • Prerequisites
  • Installation
  • Quickstart
  • Prepare Data
  • Prepare and Validate
  • Download from Hugging Face
  • Prompt Config
  • Configure Agents
  • Integrate Existing Agents
  • Drive a Remote Agent
  • Agent Skills
  • Configure Models
  • OpenAI
  • Azure OpenAI
  • Inference Providers
  • Model-call capture
  • vLLM
  • Local vLLM
  • Local vLLM Proxy
  • Anthropic Messages
  • Switchyard
  • Build Verifiers
  • Verification Patterns
  • Equivalence Match
  • Execution and State Match
  • LLM-as-Judge
  • Multi-Reward Verification
  • Evaluate
  • Benchmarks
  • Compare Harnesses
  • Browse Environments
  • Aggregate Metrics
  • Rollout Health Checks
  • Diagnose Results
  • Training Tutorials
  • NeMo RL
  • About Workplace Assistant
  • Gym Configuration
  • Multi-Node Training
  • NeMo RL Configuration
  • Setup
  • Single Node Training
  • Unsloth
  • Multi-Environment Training
  • Training with VeRL
  • Offline Training (SFT/DPO)
  • Training on External Agent Harnesses
  • Evaluation Tutorials
  • Evaluate EvalPlus
  • BLADE Analysis Skill
  • Reverify Rollouts
  • Build Environments
  • Resources Server APIs
  • Single-Step Environment
  • Multi-Step Environment
  • Stateful Environment
  • MCP Resources Server
  • Real-World Environment
  • Generating Training Data
  • Resources Server Implementation
  • Integrate external libraries
  • Model Recipes
  • Nemotron 3 Nano
  • Nemotron 3 Super
  • Infrastructure
  • Deployment Topology
  • Sandbox API
  • OpenSandbox Provider
  • Apptainer Provider
  • E2B Provider
  • OpenSandbox Best Practices
  • Debugging OpenSandbox Sandboxes
  • Docker Provider
  • ECS Fargate
  • Adding a Sandbox Provider
  • Engineering Notes
  • aiohttp vs httpx
  • Responses API
  • SWE RL Case Study
  • System Design
  • Observability
  • Configuration
  • Distributed Tracing
  • Span Groups
  • Metrics
  • Extending
  • Reference
  • Configuration
  • RL Framework Compatibility
  • CLI Commands
  • FAQ
  • Trajectory capability matrix
  • Troubleshooting
  • Configuration Errors
  • Contribute
  • Development Setup
  • Environments
  • Add a benchmark
  • New Environment
  • Integrate RL Frameworks
  • Generation Backend
  • Integration Footprint
  • On-Policy Corrections
  • Success Criteria
  • Agent Skills
On this page
  • RL (GRPO)
  • Multi-Environment Training
  • SFT & DPO
Tutorials

Training Tutorials

||View as Markdown|

We have hands-on tutorials with supported training frameworks to help you train with NeMo Gym environments. If you’re interested in integrating another training framework, see the Training Framework Integration Guide.

See Training for a refresher on when to use GRPO, SFT, or DPO.

RL (GRPO)

NeMo RL

Tutorial-series: GRPO training to improve multi-step tool calling on the Workplace Assistant environment, scaling from single-node to multi-node training.

nemo rlgrpo3-5 hours
Unsloth

Example GRPO training on instruction following and reasoning environments.

unslothsingle-gpu30 min
External Agent Harnesses

Train on rollouts from a harness that drives its own model calls and returns no token ids, such as the Claude Code CLI.

token captureagent harnesses
VeRL

Example DAPO training on math and agentic environments using VeRL, with single and multi-environment support.

verldapomulti-node1 hour

Multi-Environment Training

Multi-Environment Training

Run multiple training environments simultaneously for rollout collection.

multi-environmentmulti-verifier

SFT & DPO

Offline Training with Rollouts

Transform rollouts into training data for supervised fine-tuning (SFT) and direct preference optimization (DPO).

sftdpo
Previous
Diagnose Results
Next
NeMo RL
NVIDIANVIDIA
Developer-friendly docs for your API
Privacy Policy | Your Privacy Choices | Terms of Service | Accessibility | Corporate Policies | Product Security | Contact

Copyright © 2026, NVIDIA Corporation.