Thermometer: Towards Universal Calibration for Large Language Models

Maohao Shen, Subhro Das, Kristjan Greenewald, Prasanna Sattigeri, Gregory W. Wornell, Soumya Ghosh
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:44687-44711, 2024.

Abstract

We consider the issue of calibration in large language models (LLM). Recent studies have found that common interventions such as instruction tuning often result in poorly calibrated LLMs. Although calibration is well-explored in traditional applications, calibrating LLMs is uniquely challenging. These challenges stem as much from the severe computational requirements of LLMs as from their versatility, which allows them to be applied to diverse tasks. Addressing these challenges, we propose THERMOMETER, a calibration approach tailored to LLMs. THERMOMETER learns an auxiliary model, given data from multiple tasks, for calibrating a LLM. It is computationally efficient, preserves the accuracy of the LLM, and produces better-calibrated responses for new tasks. Extensive empirical evaluations across various benchmarks demonstrate the effectiveness of the proposed method.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-shen24c, title = {Thermometer: Towards Universal Calibration for Large Language Models}, author = {Shen, Maohao and Das, Subhro and Greenewald, Kristjan and Sattigeri, Prasanna and Wornell, Gregory W. and Ghosh, Soumya}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {44687--44711}, year = {2024}, editor = {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine and Weller, Adrian and Oliver, Nuria and Scarlett, Jonathan and Berkenkamp, Felix}, volume = {235}, series = {Proceedings of Machine Learning Research}, month = {21--27 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v235/main/assets/shen24c/shen24c.pdf}, url = {https://proceedings.mlr.press/v235/shen24c.html}, abstract = {We consider the issue of calibration in large language models (LLM). Recent studies have found that common interventions such as instruction tuning often result in poorly calibrated LLMs. Although calibration is well-explored in traditional applications, calibrating LLMs is uniquely challenging. These challenges stem as much from the severe computational requirements of LLMs as from their versatility, which allows them to be applied to diverse tasks. Addressing these challenges, we propose THERMOMETER, a calibration approach tailored to LLMs. THERMOMETER learns an auxiliary model, given data from multiple tasks, for calibrating a LLM. It is computationally efficient, preserves the accuracy of the LLM, and produces better-calibrated responses for new tasks. Extensive empirical evaluations across various benchmarks demonstrate the effectiveness of the proposed method.} }
Endnote
%0 Conference Paper %T Thermometer: Towards Universal Calibration for Large Language Models %A Maohao Shen %A Subhro Das %A Kristjan Greenewald %A Prasanna Sattigeri %A Gregory W. Wornell %A Soumya Ghosh %B Proceedings of the 41st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2024 %E Ruslan Salakhutdinov %E Zico Kolter %E Katherine Heller %E Adrian Weller %E Nuria Oliver %E Jonathan Scarlett %E Felix Berkenkamp %F pmlr-v235-shen24c %I PMLR %P 44687--44711 %U https://proceedings.mlr.press/v235/shen24c.html %V 235 %X We consider the issue of calibration in large language models (LLM). Recent studies have found that common interventions such as instruction tuning often result in poorly calibrated LLMs. Although calibration is well-explored in traditional applications, calibrating LLMs is uniquely challenging. These challenges stem as much from the severe computational requirements of LLMs as from their versatility, which allows them to be applied to diverse tasks. Addressing these challenges, we propose THERMOMETER, a calibration approach tailored to LLMs. THERMOMETER learns an auxiliary model, given data from multiple tasks, for calibrating a LLM. It is computationally efficient, preserves the accuracy of the LLM, and produces better-calibrated responses for new tasks. Extensive empirical evaluations across various benchmarks demonstrate the effectiveness of the proposed method.
APA
Shen, M., Das, S., Greenewald, K., Sattigeri, P., Wornell, G.W. & Ghosh, S.. (2024). Thermometer: Towards Universal Calibration for Large Language Models. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:44687-44711 Available from https://proceedings.mlr.press/v235/shen24c.html.

Related Material