Lightweight Latent Verifiers for Efficient Meta-Generation Strategies

2025-04-23

AI Summary

To address the prohibitively high computational overhead of large language model (LLM) verifiers in inference-intensive tasks, this paper proposes LiLaVe—a lightweight latent-state verifier. Methodologically, LiLaVe innovatively extracts correctness signals directly from intermediate hidden layers of the base LLM, eliminating the need for costly explicit LLM-based verification. It introduces novel meta-generation strategies—including conditional self-correction and conditional majority voting—to jointly optimize both accuracy and inference efficiency of small models. Furthermore, LiLaVe integrates latent-state feature decoding, lightweight classification head training, and ensemble techniques (e.g., best-of-n and self-consistency) for end-to-end verification acceleration. Experimental results demonstrate that LiLaVe matches the accuracy of full LLM verifiers across diverse reasoning tasks while reducing verification-stage computational cost by over 90%. This yields substantial improvements in the inference efficiency of small models without compromising reliability.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Large Vision Models
• Natural Language Processing: (Large) Language Models

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Large language models for search

Abstract

Verifiers are auxiliary models that assess the correctness of outputs generated by base large language models (LLMs). They play a crucial role in many strategies for solving reasoning-intensive problems with LLMs. Typically, verifiers are LLMs themselves, often as large (or larger) than the base model they support, making them computationally expensive. In this work, we introduce a novel lightweight verification approach, LiLaVe, which reliably extracts correctness signals from the hidden states of the base LLM. A key advantage of LiLaVe is its ability to operate with only a small fraction of the computational budget required by traditional LLM-based verifiers. To demonstrate its practicality, we couple LiLaVe with popular meta-generation strategies, like best-of-n or self-consistency. Moreover, we design novel LiLaVe-based approaches, like conditional self-correction or conditional majority voting, that significantly improve both accuracy and efficiency in generation tasks with smaller LLMs. Our work demonstrates the fruitfulness of extracting latent information from the hidden states of LLMs, and opens the door to scalable and resource-efficient solutions for reasoning-intensive applications.

Problem

• Reducing computational cost of LLM verifiers
• Extracting correctness signals from base LLM hidden states
• Improving accuracy and efficiency in generation tasks

Innovation

• Lightweight verifiers extract signals from hidden states
• LiLaVe reduces computational cost significantly
• Novel strategies improve accuracy and efficiency