Sign up to save your podcasts
Or
Share REASONING GYM: Reasoning Environments for Reinforcement Learning with Verifiable Rewards
Share to email
Share to Facebook
Share to X
Share to email
Share to Facebook
Share to X
Or
REASONING GYM: Reasoning Environments for Reinforcement Learning with Verifiable Rewards
🤗 Upvotes: 52 | cs.LG, cs.AI, cs.CL
Authors:
Zafir Stojanovski, Oliver Stanley, Joe Sharratt, Richard Jones, Abdulhakeem Adefioye, Jean Kaddour, Andreas Köpf
Title:
REASONING GYM: Reasoning Environments for Reinforcement Learning with Verifiable Rewards
Arxiv:
http://arxiv.org/abs/2505.24760v1
Abstract:
We introduce Reasoning Gym (RG), a library of reasoning environments for reinforcement learning with verifiable rewards. It provides over 100 data generators and verifiers spanning multiple domains including algebra, arithmetic, computation, cognition, geometry, graph theory, logic, and various common games. Its key innovation is the ability to generate virtually infinite training data with adjustable complexity, unlike most previous reasoning datasets, which are typically fixed. This procedural generation approach allows for continuous evaluation across varying difficulty levels. Our experimental results demonstrate the efficacy of RG in both evaluating and reinforcement learning of reasoning models.
View all episodes
By Jingwen Liang, Gengyu Wang🤗 Upvotes: 52 | cs.LG, cs.AI, cs.CL
Authors:
Zafir Stojanovski, Oliver Stanley, Joe Sharratt, Richard Jones, Abdulhakeem Adefioye, Jean Kaddour, Andreas Köpf
Title:
REASONING GYM: Reasoning Environments for Reinforcement Learning with Verifiable Rewards
Arxiv:
http://arxiv.org/abs/2505.24760v1
Abstract:
We introduce Reasoning Gym (RG), a library of reasoning environments for reinforcement learning with verifiable rewards. It provides over 100 data generators and verifiers spanning multiple domains including algebra, arithmetic, computation, cognition, geometry, graph theory, logic, and various common games. Its key innovation is the ability to generate virtually infinite training data with adjustable complexity, unlike most previous reasoning datasets, which are typically fixed. This procedural generation approach allows for continuous evaluation across varying difficulty levels. Our experimental results demonstrate the efficacy of RG in both evaluating and reinforcement learning of reasoning models.