Torchforge Rl Training
Provides guidance for PyTorch-native agentic RL using torchforge, Meta's library...
Install
npx skills add Orchestra-Research/AI-Research-SKILLs --skill torchforge-rl-training -g -a claude-code -y- Repository
- Orchestra-Research/AI-Research-SKILLs14.2k
- Category
- Engineering
- Updated
- 25d ago
- License
- MIT
- Context cost
- Mediumabout 2,367 tokens
- Runs scripts
- No, instructions only
What it does
Provides guidance for PyTorch-native agentic RL using torchforge, Meta's library separating infra from algorithms. Use when you want clean RL abstractions, easy algorithm experimentation, or scalable training with Monarch and TorchTitan.
torchforge: PyTorch-Native Agentic RL Library
torchforge is Meta's PyTorch-native RL library that separates infrastructure concerns from algorithm concerns. It enables rapid RL research by letting you focus on algorithms while handling distributed training, inference, and weight sync automatically.
When to Use torchforge
Choose torchforge when you need:
- Clean separation between RL algorithms and infrastructure
- PyTorch-native abstractions (no Ray dependency)
- Easy algorithm experimentation (GRPO, DAPO, SAPO in ~100 lines)
- Scalable training with Monarch actor system
- Integration with TorchTitan for model parallelism
Consider alternatives when:
- You need production-ready stability → use miles or verl
- You want Megatron-native training → use slime
- torchforge is experimental and APIs may change
Key Features
- Algorithm isolation: Implement RL algorithms without touching infrastructure
- Scalability: From single GPU to thousands via Monarch
- Modern stack: TorchTitan (training), vLLM (inference), TorchStore (sync)
- Loss functions: GRPO, DAPO, CISPO, GSPO, SAPO built-in
