Uncertainty-Aware Decoding with Minimum Bayes Risk
This addresses the issue of unreliable text generation for users of language models, but it is incremental as it builds on existing MBR decoding with uncertainty integration.
The paper tackles the problem of language models generating undesirable outputs like hallucinations by proposing an uncertainty-aware decoding method based on Minimum Bayes Risk (MBR), which incorporates model parameter uncertainty to improve output selection and abstention decisions without overhead.
Despite their outstanding performance in the majority of scenarios, contemporary language models still occasionally generate undesirable outputs, for example, hallucinated text. While such behaviors have previously been linked to uncertainty, there is a notable lack of methods that actively consider uncertainty during text generation. In this work, we show how Minimum Bayes Risk (MBR) decoding, which selects model generations according to an expected risk, can be generalized into a principled uncertainty-aware decoding method. In short, we account for model uncertainty during decoding by incorporating a posterior over model parameters into MBR's computation of expected risk. We show that this modified expected risk is useful for both choosing outputs and deciding when to abstain from generation and can provide improvements without incurring overhead. We benchmark different methods for learning posteriors and show that performance improves with prediction diversity. We release our code publicly.