skills.rest
hhhi21g
hhhi21gCOMMUNITY
Explore all skills by this author →
GitHub

torchforge-rl-training

Provides a PyTorch-native RL library separating infrastructure from algorithms.

Updated Apr 11, 2026
Search Tags:#reinforcement learning#pytorch#torchforge#rl library#algorithm experimentation
One-click install
npx skills add https://github.com/hhhi21g/HealthCenter --skill torchforge-rl-training-hhhi21g
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: torchforge-rl-training
Source: https://github.com/hhhi21g/HealthCenter/tree/main/.codex/skills/torchforge
Command: npx skills add https://github.com/hhhi21g/HealthCenter --skill torchforge-rl-training-hhhi21g

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires torch>=2.9.0, torchtitan>=0.2.0, vllm, monarch, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill addresses the complexities of PyTorch-native Reinforcement Learning (RL) by providing a library that separates infrastructure from algorithms, simplifying RL research and experimentation.

Core Features & Use Cases

  • Algorithm Isolation: Focus on algorithms without worrying about infrastructure.
  • Scalability: Supports training on single GPUs to thousands with Monarch and TorchTitan.
  • Modern Stack: Integrates with TorchTitan for training, vLLM for inference, and TorchStore for synchronization.
  • Use Case: Ideal for researchers and developers looking to implement and experiment with RL algorithms efficiently.

Quick Start

Train a GRPO model for math reasoning using the torchforge skill with the following command:

python -m apps.grpo.main --config config/grpo_math.yaml

Frequently Asked Questions about torchforge-rl-training

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

FAQPage Schema
How do I scale PyTorch reinforcement learning from a single GPU to thousands?▼

Scale PyTorch reinforcement learning by using a library that integrates with Monarch and TorchTitan, enabling distributed training across thousands of GPUs while isolating infrastructure from algorithms.

What is the best way to separate RL algorithms from infrastructure in PyTorch?▼

Separate RL algorithms from infrastructure in PyTorch by using a native library designed for research experimentation, allowing developers to focus solely on algorithm implementation without managing underlying systems.

Does torchforge work with vLLM for reinforcement learning inference?▼

Torchforge works with vLLM for reinforcement learning inference by integrating it into the modern stack, alongside TorchTitan for training and TorchStore for synchronization, streamlining research workflows.

How do I train a GRPO model for math reasoning using PyTorch?▼

Train a GRPO model for math reasoning by running the provided main script with a specified YAML configuration file, initiating the training process natively within the PyTorch ecosystem.

Do I need PyTorch 2.9.0 to run reinforcement learning experiments with this library?▼

You need PyTorch version 2.9.0 or higher, along with torchtitan 0.2.0, vLLM, and Monarch, to run reinforcement learning experiments and ensure full compatibility with the library's modern stack.

Related Skills

Explore More Software Engineering Skills→
zhuangbiaoweizhuangbiaowei
2

torchforge-rl-training

Train PyTorch-native reinforcement learning algorithms with distributed infrastructure.

Community
Advanced
shuff57shuff57

grpo-rl-training

Fine-tune language models with GRPO and TRL for verifiable tasks.

Community
Advanced
Orchestra-ResearchOrchestra-Research
11.5k

torchforge-rl-training

Facilitates PyTorch-native RL training with Monarch, TorchTitan, and vLLM for distributed execution and inference.

Official
Advanced
jonnabiojonnabio
6

torchforge-rl-training

Prototype and train PyTorch RL algorithms with GRPO, DAPO, and SAPO.

Community
Advanced