Decoding Uncertainty: The Impact of Decoding Strategies for Uncertainty Estimation in Large Language Models
This work addresses uncertainty estimation for LLM users, but it is incremental as it builds on existing decoding methods.
The study examined how different decoding strategies affect uncertainty estimation in Large Language Models, finding that Contrastive Search generally improves uncertainty estimates in preference-aligned models, but benefits vary in models with only supervised fine-tuning.
Decoding strategies manipulate the probability distribution underlying the output of a language model and can therefore affect both generation quality and its uncertainty. In this study, we investigate the impact of decoding strategies on uncertainty estimation in Large Language Models (LLMs). Our experiments show that Contrastive Search, which mitigates repetition, yields better uncertainty estimates on average across a range of preference-aligned LLMs. In contrast, the benefits of these strategies sometimes diverge when the model is only post-trained with supervised fine-tuning, i.e. without explicit alignment.