Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2406.12319 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915249133715456 |
|---|---|
| author | Jeong, Hawon Park, ChaeHun Hong, Jimin Lee, Hojoon Choo, Jaegul |
| author_facet | Jeong, Hawon Park, ChaeHun Hong, Jimin Lee, Hojoon Choo, Jaegul |
| contents | As large language models (LLMs) are increasingly used as evaluators for natural language generation tasks, ensuring unbiased assessments is essential. However, LLM evaluators often display biased preferences, such as favoring verbosity and authoritative tones. Our empirical analysis reveals that these biases are exacerbated in pairwise evaluation, where LLMs directly compare two outputs and easily prioritize superficial attributes. In contrast, pointwise evaluation, which assesses outputs independently, is less susceptible to such bias because each output is judged in isolation. To address the limitations of the pairwise evaluation, we introduce a novel evaluation method, PRePair, which integrates pointwise reasoning within a pairwise framework. PRePair effectively alleviates biased preference, improving performance on the adversarial benchmark (LLMBar) while outperforming pointwise evaluation on the standard benchmark (MT-Bench). |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2406_12319 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | The Comparative Trap: Pairwise Comparisons Amplifies Biased Preferences of LLM Evaluators Jeong, Hawon Park, ChaeHun Hong, Jimin Lee, Hojoon Choo, Jaegul Computation and Language As large language models (LLMs) are increasingly used as evaluators for natural language generation tasks, ensuring unbiased assessments is essential. However, LLM evaluators often display biased preferences, such as favoring verbosity and authoritative tones. Our empirical analysis reveals that these biases are exacerbated in pairwise evaluation, where LLMs directly compare two outputs and easily prioritize superficial attributes. In contrast, pointwise evaluation, which assesses outputs independently, is less susceptible to such bias because each output is judged in isolation. To address the limitations of the pairwise evaluation, we introduce a novel evaluation method, PRePair, which integrates pointwise reasoning within a pairwise framework. PRePair effectively alleviates biased preference, improving performance on the adversarial benchmark (LLMBar) while outperforming pointwise evaluation on the standard benchmark (MT-Bench). |
| title | The Comparative Trap: Pairwise Comparisons Amplifies Biased Preferences of LLM Evaluators |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2406.12319 |