Inferring from Logits: Exploring Best Practices for Decoding-Free Generative Candidate Selection

TL;DR

This paper systematically evaluates decoding-free candidate selection methods for generative language models across diverse tasks and models, providing insights to improve task-level output prediction efficiency.

Contribution

It offers a comprehensive evaluation of decoding-free candidate selection approaches on multiple tasks and models, highlighting their effectiveness and guiding future model design.

Findings

Decoding-free methods vary in accuracy depending on task and model.

Some methods perform well on small candidate pools but struggle with large ones.

Insights inform better candidate selection strategies for generative models.

Abstract

Generative Language Models rely on autoregressive decoding to produce the output sequence token by token. Many tasks such as preference optimization, require the model to produce task-level output consisting of multiple tokens directly by selecting candidates from a pool as predictions. Determining a task-level prediction from candidates using the ordinary token-level decoding mechanism is constrained by time-consuming decoding and interrupted gradients by discrete token selection. Existing works have been using decoding-free candidate selection methods to obtain candidate probability from initial output logits over vocabulary. Though these estimation methods are widely used, they are not systematically evaluated, especially on end tasks. We introduce an evaluation of a comprehensive collection of decoding-free candidate selection approaches on a comprehensive set of tasks, including five multiple-choice QA tasks with a small candidate pool and four clinical decision tasks with a massive amount of candidates, some with 10k+ options. We evaluate the estimation methods paired with a wide spectrum of foundation LMs covering different architectures, sizes and training paradigms. The results and insights from our analysis inform the future model design.