Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2501.06248 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916629686779904 |
|---|---|
| author | Maura-Rivero, Roberto-Rafael Nagpal, Chirag Patel, Roma Visin, Francesco |
| author_facet | Maura-Rivero, Roberto-Rafael Nagpal, Chirag Patel, Roma Visin, Francesco |
| contents | Current methods that train large language models (LLMs) with reinforcement learning feedback, often resort to averaging outputs of multiple rewards functions during training. This overlooks crucial aspects of individual reward dimensions and inter-reward dependencies that can lead to sub-optimal outcomes in generations. In this work, we show how linear aggregation of rewards exhibits some vulnerabilities that can lead to undesired properties of generated text. We then propose a transformation of reward functions inspired by economic theory of utility functions (specifically Inada conditions), that enhances sensitivity to low reward values while diminishing sensitivity to already high values. We compare our approach to the existing baseline methods that linearly aggregate rewards and show how the Inada-inspired reward feedback is superior to traditional weighted averaging. We quantitatively and qualitatively analyse the difference in the methods, and see that models trained with Inada-transformations score as more helpful while being less harmful. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2501_06248 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Utility-inspired Reward Transformations Improve Reinforcement Learning Training of Language Models Maura-Rivero, Roberto-Rafael Nagpal, Chirag Patel, Roma Visin, Francesco Machine Learning Artificial Intelligence Computation and Language General Economics Economics Current methods that train large language models (LLMs) with reinforcement learning feedback, often resort to averaging outputs of multiple rewards functions during training. This overlooks crucial aspects of individual reward dimensions and inter-reward dependencies that can lead to sub-optimal outcomes in generations. In this work, we show how linear aggregation of rewards exhibits some vulnerabilities that can lead to undesired properties of generated text. We then propose a transformation of reward functions inspired by economic theory of utility functions (specifically Inada conditions), that enhances sensitivity to low reward values while diminishing sensitivity to already high values. We compare our approach to the existing baseline methods that linearly aggregate rewards and show how the Inada-inspired reward feedback is superior to traditional weighted averaging. We quantitatively and qualitatively analyse the difference in the methods, and see that models trained with Inada-transformations score as more helpful while being less harmful. |
| title | Utility-inspired Reward Transformations Improve Reinforcement Learning Training of Language Models |
| topic | Machine Learning Artificial Intelligence Computation and Language General Economics Economics |
| url | https://arxiv.org/abs/2501.06248 |