Your Finetuned Large Language Model is Already a Powerful Out-of-distribution Detector

Andi Zhang, Tim Z. Xiao, Weiyang Liu, Robert Bamler, Damon Wischik
Proceedings of The 28th International Conference on Artificial Intelligence and Statistics, PMLR 258:2701-2709, 2025.

Abstract

We revisit the likelihood ratio between a pretrained large language model (LLM) and its finetuned variant as a criterion for out-of-distribution (OOD) detection. The intuition behind such a criterion is that, the pretrained LLM has the prior knowledge about OOD data due to its large amount of training data, and once finetuned with the in-distribution data, the LLM has sufficient knowledge to distinguish their difference. Leveraging the power of LLMs, we show that, the likelihood ratio can serve as an effective OOD detection criterion. Moreover, we apply the proposed LLM-based likelihood ratio to detect OOD questions in question-answering (QA) systems, which can be used to improve the performance of specialized LLMs for general questions. Given that likelihood can be easily obtained by the loss functions within contemporary neural network frameworks, it is straightforward to implement this approach in practice with \textbf{three lines} of code. Since both the pretrained LLMs and its various finetuned models are widely available from online platforms such as Hugging Face, our proposed criterion can be effortlessly incorporated for OOD detection without the need for further training. We conduct comprehensive evaluation across on multiple settings, including far OOD, near OOD, spam detection, and QA scenarios, to demonstrate the effectiveness of the method. Code can be found at \url{https://github.com/andiac/LLMOODratio}

Cite this Paper

BibTeX
@InProceedings{pmlr-v258-zhang25h, title = {Your Finetuned Large Language Model is Already a Powerful Out-of-distribution Detector}, author = {Zhang, Andi and Xiao, Tim Z. and Liu, Weiyang and Bamler, Robert and Wischik, Damon}, booktitle = {Proceedings of The 28th International Conference on Artificial Intelligence and Statistics}, pages = {2701--2709}, year = {2025}, editor = {Li, Yingzhen and Mandt, Stephan and Agrawal, Shipra and Khan, Emtiyaz}, volume = {258}, series = {Proceedings of Machine Learning Research}, month = {03--05 May}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v258/main/assets/zhang25h/zhang25h.pdf}, url = {https://proceedings.mlr.press/v258/zhang25h.html}, abstract = {We revisit the likelihood ratio between a pretrained large language model (LLM) and its finetuned variant as a criterion for out-of-distribution (OOD) detection. The intuition behind such a criterion is that, the pretrained LLM has the prior knowledge about OOD data due to its large amount of training data, and once finetuned with the in-distribution data, the LLM has sufficient knowledge to distinguish their difference. Leveraging the power of LLMs, we show that, the likelihood ratio can serve as an effective OOD detection criterion. Moreover, we apply the proposed LLM-based likelihood ratio to detect OOD questions in question-answering (QA) systems, which can be used to improve the performance of specialized LLMs for general questions. Given that likelihood can be easily obtained by the loss functions within contemporary neural network frameworks, it is straightforward to implement this approach in practice with \textbf{three lines} of code. Since both the pretrained LLMs and its various finetuned models are widely available from online platforms such as Hugging Face, our proposed criterion can be effortlessly incorporated for OOD detection without the need for further training. We conduct comprehensive evaluation across on multiple settings, including far OOD, near OOD, spam detection, and QA scenarios, to demonstrate the effectiveness of the method. Code can be found at \url{https://github.com/andiac/LLMOODratio}} }
Endnote
%0 Conference Paper %T Your Finetuned Large Language Model is Already a Powerful Out-of-distribution Detector %A Andi Zhang %A Tim Z. Xiao %A Weiyang Liu %A Robert Bamler %A Damon Wischik %B Proceedings of The 28th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2025 %E Yingzhen Li %E Stephan Mandt %E Shipra Agrawal %E Emtiyaz Khan %F pmlr-v258-zhang25h %I PMLR %P 2701--2709 %U https://proceedings.mlr.press/v258/zhang25h.html %V 258 %X We revisit the likelihood ratio between a pretrained large language model (LLM) and its finetuned variant as a criterion for out-of-distribution (OOD) detection. The intuition behind such a criterion is that, the pretrained LLM has the prior knowledge about OOD data due to its large amount of training data, and once finetuned with the in-distribution data, the LLM has sufficient knowledge to distinguish their difference. Leveraging the power of LLMs, we show that, the likelihood ratio can serve as an effective OOD detection criterion. Moreover, we apply the proposed LLM-based likelihood ratio to detect OOD questions in question-answering (QA) systems, which can be used to improve the performance of specialized LLMs for general questions. Given that likelihood can be easily obtained by the loss functions within contemporary neural network frameworks, it is straightforward to implement this approach in practice with \textbf{three lines} of code. Since both the pretrained LLMs and its various finetuned models are widely available from online platforms such as Hugging Face, our proposed criterion can be effortlessly incorporated for OOD detection without the need for further training. We conduct comprehensive evaluation across on multiple settings, including far OOD, near OOD, spam detection, and QA scenarios, to demonstrate the effectiveness of the method. Code can be found at \url{https://github.com/andiac/LLMOODratio}
APA
Zhang, A., Xiao, T.Z., Liu, W., Bamler, R. & Wischik, D.. (2025). Your Finetuned Large Language Model is Already a Powerful Out-of-distribution Detector. Proceedings of The 28th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 258:2701-2709 Available from https://proceedings.mlr.press/v258/zhang25h.html.

Related Material