Benchmarking Uncertainty Quantification Methods for Large Language Models with LM-Polygraph
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909665962491904 |
|---|---|
| author | Vashurin, Roman Fadeeva, Ekaterina Vazhentsev, Artem Rvanova, Lyudmila Tsvigun, Akim Vasilev, Daniil Xing, Rui Sadallah, Abdelrahman Boda Grishchenkov, Kirill Petrakov, Sergey Panchenko, Alexander Baldwin, Timothy Nakov, Preslav Panov, Maxim Shelmanov, Artem |
| author_facet | Vashurin, Roman Fadeeva, Ekaterina Vazhentsev, Artem Rvanova, Lyudmila Tsvigun, Akim Vasilev, Daniil Xing, Rui Sadallah, Abdelrahman Boda Grishchenkov, Kirill Petrakov, Sergey Panchenko, Alexander Baldwin, Timothy Nakov, Preslav Panov, Maxim Shelmanov, Artem |
| contents | The rapid proliferation of large language models (LLMs) has stimulated researchers to seek effective and efficient approaches to deal with LLM hallucinations and low-quality outputs. Uncertainty quantification (UQ) is a key element of machine learning applications in dealing with such challenges. However, research to date on UQ for LLMs has been fragmented in terms of techniques and evaluation methodologies. In this work, we address this issue by introducing a novel benchmark that implements a collection of state-of-the-art UQ baselines and offers an environment for controllable and consistent evaluation of novel UQ techniques over various text generation tasks. Our benchmark also supports the assessment of confidence normalization methods in terms of their ability to provide interpretable scores. Using our benchmark, we conduct a large-scale empirical investigation of UQ and normalization techniques across eleven tasks, identifying the most effective approaches. Code: https://github.com/IINemo/lm-polygraph Benchmark: https://huggingface.co/LM-Polygraph |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2406_15627 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Benchmarking Uncertainty Quantification Methods for Large Language Models with LM-Polygraph Vashurin, Roman Fadeeva, Ekaterina Vazhentsev, Artem Rvanova, Lyudmila Tsvigun, Akim Vasilev, Daniil Xing, Rui Sadallah, Abdelrahman Boda Grishchenkov, Kirill Petrakov, Sergey Panchenko, Alexander Baldwin, Timothy Nakov, Preslav Panov, Maxim Shelmanov, Artem Computation and Language Machine Learning The rapid proliferation of large language models (LLMs) has stimulated researchers to seek effective and efficient approaches to deal with LLM hallucinations and low-quality outputs. Uncertainty quantification (UQ) is a key element of machine learning applications in dealing with such challenges. However, research to date on UQ for LLMs has been fragmented in terms of techniques and evaluation methodologies. In this work, we address this issue by introducing a novel benchmark that implements a collection of state-of-the-art UQ baselines and offers an environment for controllable and consistent evaluation of novel UQ techniques over various text generation tasks. Our benchmark also supports the assessment of confidence normalization methods in terms of their ability to provide interpretable scores. Using our benchmark, we conduct a large-scale empirical investigation of UQ and normalization techniques across eleven tasks, identifying the most effective approaches. Code: https://github.com/IINemo/lm-polygraph Benchmark: https://huggingface.co/LM-Polygraph |
| title | Benchmarking Uncertainty Quantification Methods for Large Language Models with LM-Polygraph |
| topic | Computation and Language Machine Learning |
| url | https://arxiv.org/abs/2406.15627 |