# Comparison of Diverse Decoding Methods from Conditional Language Models

## TL;DR

This paper surveys various decoding strategies for conditional language models, emphasizing methods to generate diverse high-quality outputs, and proposes over-sampling combined with filtering as an effective approach.

## Contribution

It provides an extensive survey of diverse decoding methods and introduces a simple yet effective over-sampling and filtering technique to enhance diversity without losing quality.

## Key findings

- Over-sampling and filtering improves diversity effectively.
- Standard beam search focuses on likelihood, not diversity.
- Diverse decoding enhances re-ranking and candidate selection.

## Abstract

While conditional language models have greatly improved in their ability to output high-quality natural language, many NLP applications benefit from being able to generate a diverse set of candidate sequences. Diverse decoding strategies aim to, within a given-sized candidate list, cover as much of the space of high-quality outputs as possible, leading to improvements for tasks that re-rank and combine candidate outputs. Standard decoding methods, such as beam search, optimize for generating high likelihood sequences rather than diverse ones, though recent work has focused on increasing diversity in these methods. In this work, we perform an extensive survey of decoding-time strategies for generating diverse outputs from conditional language models. We also show how diversity can be improved without sacrificing quality by over-sampling additional candidates, then filtering to the desired number.