Graph-based Confidence Calibration for Large Language Models
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_ | 1866912386178351104 |
|---|---|
| author | Li, Yukun Wang, Sijia Huang, Lifu Liu, Li-Ping |
| author_facet | Li, Yukun Wang, Sijia Huang, Lifu Liu, Li-Ping |
| contents | Reliable confidence estimation is essential for enhancing the trustworthiness of large language models (LLMs), especially in high-stakes scenarios. Despite its importance, accurately estimating confidence in LLM responses remains a significant challenge. In this work, we propose using an auxiliary learning model to assess response correctness based on the self-consistency of multiple outputs generated by the LLM. Our method builds a consistency graph to represent the agreement among multiple responses and uses a graph neural network (GNN) to estimate the likelihood that each response is correct. Experiments demonstrate that this method has strong calibration performance on various benchmark datasets and generalizes well to out-of-domain cases. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2411_02454 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Graph-based Confidence Calibration for Large Language Models Li, Yukun Wang, Sijia Huang, Lifu Liu, Li-Ping Computation and Language Artificial Intelligence Information Retrieval Machine Learning Reliable confidence estimation is essential for enhancing the trustworthiness of large language models (LLMs), especially in high-stakes scenarios. Despite its importance, accurately estimating confidence in LLM responses remains a significant challenge. In this work, we propose using an auxiliary learning model to assess response correctness based on the self-consistency of multiple outputs generated by the LLM. Our method builds a consistency graph to represent the agreement among multiple responses and uses a graph neural network (GNN) to estimate the likelihood that each response is correct. Experiments demonstrate that this method has strong calibration performance on various benchmark datasets and generalizes well to out-of-domain cases. |
| title | Graph-based Confidence Calibration for Large Language Models |
| topic | Computation and Language Artificial Intelligence Information Retrieval Machine Learning |
| url | https://arxiv.org/abs/2411.02454 |