Regression-aware Inference with LLMs
This work addresses a specific bottleneck in LLM inference for regression and scoring tasks, offering an incremental improvement over existing methods.
The paper tackles the problem of suboptimal inference strategies in large language models for regression and scoring tasks, proposing alternate inference strategies that estimate the Bayes-optimal solution in closed-form, resulting in significant improvements over baselines across datasets and models.
Large language models (LLMs) have shown strong results on a range of applications, including regression and scoring tasks. Typically, one obtains outputs from an LLM via autoregressive sampling from the model's output distribution. We show that this inference strategy can be sub-optimal for common regression and scoring evaluation metrics. As a remedy, we build on prior work on Minimum Bayes Risk decoding, and propose alternate inference strategies that estimate the Bayes-optimal solution for regression and scoring metrics in closed-form from sampled responses. We show that our proposal significantly improves over baselines across datasets and models.