GIFT: Games as Informal Training for Generalizable LLMs

TL;DR

This paper introduces a novel approach where Large Language Models learn through interactive games, enhancing their generalizable intelligence and strategic reasoning by using a nested training framework with reinforcement learning.

Contribution

It proposes treating games as a primary environment for informal learning and introduces a nested training framework to improve multi-task learning in LLMs.

Findings

Game-based informal learning improves generalization across benchmarks.

Nested training framework prevents task interference.

Reinforcement learning with games enhances strategic and social reasoning.

Abstract

While Large Language Models (LLMs) have achieved remarkable success in formal learning tasks such as mathematics and code generation, they still struggle with the "practical wisdom" and generalizable intelligence, such as strategic creativity and social reasoning, that characterize human cognition. This gap arises from a lack of informal learning, which thrives on interactive feedback rather than goal-oriented instruction. In this paper, we propose treating Games as a primary environment for LLM informal learning, leveraging their intrinsic reward signals and abstracted complexity to cultivate diverse competencies. To address the performance degradation observed in multi-task learning, we introduce a Nested Training Framework. Unlike naive task mixing optimizing an implicit "OR" objective, our framework employs sequential task composition to enforce an explicit "AND" objective, compelling the model to master multiple abilities simultaneously to achieve maximal rewards. Using GRPO-based reinforcement learning across Matrix Games, TicTacToe, and Who's the Spy games, we demonstrate that integrating game-based informal learning not only prevents task interference but also significantly bolsters the model's generalization across broad ability-oriented benchmarks. The framework and implementation are publicly available.