GIFT: Games as Informal Training for Generalizable LLMs
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918279192248320 |
|---|---|
| author | Lyu, Nuoyan Xu, Bingbing Meng, Weihao Yuan, Yige Zhang, Yang Huang, Zhiyong Chua, Tat-Seng Shen, Huawei |
| author_facet | Lyu, Nuoyan Xu, Bingbing Meng, Weihao Yuan, Yige Zhang, Yang Huang, Zhiyong Chua, Tat-Seng Shen, Huawei |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_05633 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | GIFT: Games as Informal Training for Generalizable LLMs Lyu, Nuoyan Xu, Bingbing Meng, Weihao Yuan, Yige Zhang, Yang Huang, Zhiyong Chua, Tat-Seng Shen, Huawei Computation and Language 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. |
| title | GIFT: Games as Informal Training for Generalizable LLMs |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2601.05633 |