torchforge-rl-training
orchestra-research890 installs
Build RL training code apart from infrastructure.
What it does
Guides the implementation and training of reinforcement-learning models with torchforge, including custom losses and distributed configurations.
When to use it
Use when developing RL algorithms with separate infrastructure, experimenting with losses, or scaling training through Monarch and TorchTitan.
How to use it
It changes how the agent implements RL training configurations, reward functions, and custom losses in the project.
What you provide
- An existing project
Uses
Latest review
What you need · 4
The documented environment uses Python 3.12.
The installation procedure uses conda to create and activate the training environment.
Access · 5
This skill
HuggingFace Hub
Read
Reads models and training data
W&B
Read
Reads W&B training curves