Snowflake Research

Arctic RL Recipes Overview

RecipesAgent View (Markdown)

Overview

Arctic RL Recipes provides end-to-end training recipes for reinforcement learning models using Arctic RL, with examples ranging from simple single-GPU setups to specialized tasks like text-to-SQL and long-context question answering.

Arctic RL Recipes Overview

Arctic RL Recipes offers end-to-end training recipes for models using Arctic RL.

Available Recipes

The following recipes are available to get started with Arctic RL:

Simple Single-GPU (GSM8K)

A minimal end-to-end training loop designed for single GPU setups without requiring Ray or a hostfile configuration. This recipe is ideal for getting started and understanding the basics of Arctic RL.

  • Location: simple
  • Best for: Beginners, testing, and single-GPU environments

Txt2SQL

A recipe for training models on text-to-SQL tasks, enabling models to translate natural language queries into SQL statements.

Long-Context QA

A recipe for training models on long-context question answering tasks, suitable for models that need to process and reason over extended text passages.