Quasi-random Multi-Sample Inference for Large Language Models
This work addresses the need for more efficient and diverse sampling in LLMs for tasks like reasoning and translation, offering an incremental improvement over existing methods.
The study tackled the problem of limited parallelizability and diversity in multi-sample decoding for large language models by introducing arithmetic sampling, which improved reasoning accuracy by 3-5% on GSM8K and translation scores by 0.45-0.89% on WMT19 tasks without significant computational overhead.
Large language models (LLMs) are often equipped with multi-sample decoding strategies. An LLM implicitly defines an arithmetic code book, facilitating efficient and embarrassingly parallelizable \textbf{arithmetic sampling} to produce multiple samples using quasi-random codes. Traditional text generation methods, such as beam search and sampling-based techniques, have notable limitations: they lack parallelizability or diversity of sampled sequences. This study explores the potential of arithmetic sampling, contrasting it with ancestral sampling across two decoding tasks that employ multi-sample inference: chain-of-thought reasoning with self-consistency and machine translation with minimum Bayes risk decoding. Our results demonstrate that arithmetic sampling produces more diverse samples, significantly improving reasoning and translation performance as the sample size increases. We observe a $\mathbf{3\text{-}5\%}$ point increase in accuracy on the GSM8K dataset and a $\mathbf{0.45\text{-}0.89\%}$ point increment in COMET score for WMT19 tasks using arithmetic sampling without any significant computational overhead.