When an LLM is apprehensive about its answers -- and when its uncertainty is justified
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866913716480507904 |
|---|---|
| author | Sychev, Petr Goncharov, Andrey Vyazhev, Daniil Khalafyan, Edvard Zaytsev, Alexey |
| author_facet | Sychev, Petr Goncharov, Andrey Vyazhev, Daniil Khalafyan, Edvard Zaytsev, Alexey |
| contents | Uncertainty estimation is crucial for evaluating Large Language Models (LLMs), particularly in high-stakes domains where incorrect answers result in significant consequences. Numerous approaches consider this problem, while focusing on a specific type of uncertainty, ignoring others. We investigate what estimates, specifically token-wise entropy and model-as-judge (MASJ), would work for multiple-choice question-answering tasks for different question topics. Our experiments consider three LLMs: Phi-4, Mistral, and Qwen of different sizes from 1.5B to 72B and $14$ topics. While MASJ performs similarly to a random error predictor, the response entropy predicts model error in knowledge-dependent domains and serves as an effective indicator of question difficulty: for biology ROC AUC is $0.73$. This correlation vanishes for the reasoning-dependent domain: for math questions ROC-AUC is $0.55$. More principally, we found out that the entropy measure required a reasoning amount. Thus, data-uncertainty related entropy should be integrated within uncertainty estimates frameworks, while MASJ requires refinement. Moreover, existing MMLU-Pro samples are biased, and should balance required amount of reasoning for different subdomains to provide a more fair assessment of LLMs performance. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_01688 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | When an LLM is apprehensive about its answers -- and when its uncertainty is justified Sychev, Petr Goncharov, Andrey Vyazhev, Daniil Khalafyan, Edvard Zaytsev, Alexey Computation and Language Machine Learning Uncertainty estimation is crucial for evaluating Large Language Models (LLMs), particularly in high-stakes domains where incorrect answers result in significant consequences. Numerous approaches consider this problem, while focusing on a specific type of uncertainty, ignoring others. We investigate what estimates, specifically token-wise entropy and model-as-judge (MASJ), would work for multiple-choice question-answering tasks for different question topics. Our experiments consider three LLMs: Phi-4, Mistral, and Qwen of different sizes from 1.5B to 72B and $14$ topics. While MASJ performs similarly to a random error predictor, the response entropy predicts model error in knowledge-dependent domains and serves as an effective indicator of question difficulty: for biology ROC AUC is $0.73$. This correlation vanishes for the reasoning-dependent domain: for math questions ROC-AUC is $0.55$. More principally, we found out that the entropy measure required a reasoning amount. Thus, data-uncertainty related entropy should be integrated within uncertainty estimates frameworks, while MASJ requires refinement. Moreover, existing MMLU-Pro samples are biased, and should balance required amount of reasoning for different subdomains to provide a more fair assessment of LLMs performance. |
| title | When an LLM is apprehensive about its answers -- and when its uncertainty is justified |
| topic | Computation and Language Machine Learning |
| url | https://arxiv.org/abs/2503.01688 |