Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2405.16856 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910460242034688 |
|---|---|
| author | Yang, Haoyan Wang, Yixuan Xu, Xingyin Zhang, Hanyuan Bian, Yirong |
| author_facet | Yang, Haoyan Wang, Yixuan Xu, Xingyin Zhang, Hanyuan Bian, Yirong |
| contents | The study explores mitigating overconfidence bias in LLMs to improve their reliability. We introduce a knowledge transfer (KT) method utilizing chain of thoughts, where "big" LLMs impart knowledge to "small" LLMs via detailed, sequential reasoning paths. This method uses advanced reasoning of larger models to fine-tune smaller models, enabling them to produce more accurate predictions with calibrated confidence. Experimental evaluation using multiple-choice questions and sentiment analysis across diverse datasets demonstrated the KT method's superiority over the vanilla and question-answer pair (QA) fine-tuning methods. The most significant improvement in three key metrics, where the KT method outperformed the vanilla and QA methods by an average of 55.3% and 43.1%, respectively. These findings underscore the KT method's potential in enhancing model trustworthiness and accuracy, offering precise outputs with well-matched confidence levels across various contexts. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2405_16856 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Can We Trust LLMs? Mitigate Overconfidence Bias in LLMs through Knowledge Transfer Yang, Haoyan Wang, Yixuan Xu, Xingyin Zhang, Hanyuan Bian, Yirong Computation and Language The study explores mitigating overconfidence bias in LLMs to improve their reliability. We introduce a knowledge transfer (KT) method utilizing chain of thoughts, where "big" LLMs impart knowledge to "small" LLMs via detailed, sequential reasoning paths. This method uses advanced reasoning of larger models to fine-tune smaller models, enabling them to produce more accurate predictions with calibrated confidence. Experimental evaluation using multiple-choice questions and sentiment analysis across diverse datasets demonstrated the KT method's superiority over the vanilla and question-answer pair (QA) fine-tuning methods. The most significant improvement in three key metrics, where the KT method outperformed the vanilla and QA methods by an average of 55.3% and 43.1%, respectively. These findings underscore the KT method's potential in enhancing model trustworthiness and accuracy, offering precise outputs with well-matched confidence levels across various contexts. |
| title | Can We Trust LLMs? Mitigate Overconfidence Bias in LLMs through Knowledge Transfer |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2405.16856 |