Optimizing Alignment with Less: Leveraging Data Augmentation for Personalized Evaluation
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866929622558441472 |
|---|---|
| author | Seraj, Javad Mohajeri, Mohammad Mahdi Dousti, Mohammad Javad Ahmadabadi, Majid Nili |
| author_facet | Seraj, Javad Mohajeri, Mohammad Mahdi Dousti, Mohammad Javad Ahmadabadi, Majid Nili |
| contents | Automatic evaluation by large language models (LLMs) is a prominent topic today; however, judgment and evaluation tasks are often subjective and influenced by various factors, making adaptation challenging. While many studies demonstrate the capabilities of state-of-the-art proprietary LLMs in comparison to human evaluators, they often struggle to adapt to reference evaluators over time, a requirement for achieving personalized judgment. Additionally, numerous works have attempted to apply open LLMs as judges or evaluators, but these efforts frequently overlook the limitations of working with scarce data. Personalized judgment is inherently associated with limited data scenarios, which are common in many real-world problems. Our work aims to present a data augmentation technique to select a more effective sample from limited data in order to align an open LLM with human preference. Our work achieves approximately 7% improvements in Pearson correlation with a reference judge over the baseline,and 30% improvement over the base model (Llama3.1-8B-Instruct) in the mathematical reasoning evaluation task. demonstrating that augmenting selecting more effective preference data enables our approach to surpass baseline methods. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2412_07429 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Optimizing Alignment with Less: Leveraging Data Augmentation for Personalized Evaluation Seraj, Javad Mohajeri, Mohammad Mahdi Dousti, Mohammad Javad Ahmadabadi, Majid Nili Computation and Language Artificial Intelligence Automatic evaluation by large language models (LLMs) is a prominent topic today; however, judgment and evaluation tasks are often subjective and influenced by various factors, making adaptation challenging. While many studies demonstrate the capabilities of state-of-the-art proprietary LLMs in comparison to human evaluators, they often struggle to adapt to reference evaluators over time, a requirement for achieving personalized judgment. Additionally, numerous works have attempted to apply open LLMs as judges or evaluators, but these efforts frequently overlook the limitations of working with scarce data. Personalized judgment is inherently associated with limited data scenarios, which are common in many real-world problems. Our work aims to present a data augmentation technique to select a more effective sample from limited data in order to align an open LLM with human preference. Our work achieves approximately 7% improvements in Pearson correlation with a reference judge over the baseline,and 30% improvement over the base model (Llama3.1-8B-Instruct) in the mathematical reasoning evaluation task. demonstrating that augmenting selecting more effective preference data enables our approach to surpass baseline methods. |
| title | Optimizing Alignment with Less: Leveraging Data Augmentation for Personalized Evaluation |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2412.07429 |