Aligning LLMs with Individual Preferences via Interaction
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866913613003882496 |
|---|---|
| author | Wu, Shujin Fung, May Qian, Cheng Kim, Jeonghwan Hakkani-Tur, Dilek Ji, Heng |
| author_facet | Wu, Shujin Fung, May Qian, Cheng Kim, Jeonghwan Hakkani-Tur, Dilek Ji, Heng |
| contents | As large language models (LLMs) demonstrate increasingly advanced capabilities, aligning their behaviors with human values and preferences becomes crucial for their wide adoption. While previous research focuses on general alignment to principles such as helpfulness, harmlessness, and honesty, the need to account for individual and diverse preferences has been largely overlooked, potentially undermining customized human experiences. To address this gap, we train LLMs that can ''interact to align'', essentially cultivating the meta-skill of LLMs to implicitly infer the unspoken personalized preferences of the current user through multi-turn conversations, and then dynamically align their following behaviors and responses to these inferred preferences. Our approach involves establishing a diverse pool of 3,310 distinct user personas by initially creating seed examples, which are then expanded through iterative self-generation and filtering. Guided by distinct user personas, we leverage multi-LLM collaboration to develop a multi-turn preference dataset containing 3K+ multi-turn conversations in tree structures. Finally, we apply supervised fine-tuning and reinforcement learning to enhance LLMs using this dataset. For evaluation, we establish the ALOE (ALign With CustOmized PrEferences) benchmark, consisting of 100 carefully selected examples and well-designed metrics to measure the customized alignment performance during conversations. Experimental results demonstrate the effectiveness of our method in enabling dynamic, personalized alignment via interaction. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2410_03642 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Aligning LLMs with Individual Preferences via Interaction Wu, Shujin Fung, May Qian, Cheng Kim, Jeonghwan Hakkani-Tur, Dilek Ji, Heng Computation and Language Artificial Intelligence Human-Computer Interaction As large language models (LLMs) demonstrate increasingly advanced capabilities, aligning their behaviors with human values and preferences becomes crucial for their wide adoption. While previous research focuses on general alignment to principles such as helpfulness, harmlessness, and honesty, the need to account for individual and diverse preferences has been largely overlooked, potentially undermining customized human experiences. To address this gap, we train LLMs that can ''interact to align'', essentially cultivating the meta-skill of LLMs to implicitly infer the unspoken personalized preferences of the current user through multi-turn conversations, and then dynamically align their following behaviors and responses to these inferred preferences. Our approach involves establishing a diverse pool of 3,310 distinct user personas by initially creating seed examples, which are then expanded through iterative self-generation and filtering. Guided by distinct user personas, we leverage multi-LLM collaboration to develop a multi-turn preference dataset containing 3K+ multi-turn conversations in tree structures. Finally, we apply supervised fine-tuning and reinforcement learning to enhance LLMs using this dataset. For evaluation, we establish the ALOE (ALign With CustOmized PrEferences) benchmark, consisting of 100 carefully selected examples and well-designed metrics to measure the customized alignment performance during conversations. Experimental results demonstrate the effectiveness of our method in enabling dynamic, personalized alignment via interaction. |
| title | Aligning LLMs with Individual Preferences via Interaction |
| topic | Computation and Language Artificial Intelligence Human-Computer Interaction |
| url | https://arxiv.org/abs/2410.03642 |