RAVEL: Reasoning Agents for Validating and Evaluating LLM Text Synthesis

TL;DR

RAVEL introduces an agentic framework for evaluating and validating complex LLM text synthesis operations, revealing that reasoning ability significantly impacts synthesis quality more than raw generative power.

Contribution

The paper presents RAVEL, a novel framework enabling autonomous planning and execution of synthesis tasks, and introduces C3EBench, a comprehensive benchmark for detailed LLM evaluation.

Findings

Most LLMs struggle with context-understanding tasks under limited instructions.

Agentic synthesis performance is driven more by reasoning than generative capacity.

A strong reasoner can improve weaker generators' output quality.

Abstract

Large Language Models have evolved from single-round generators into long-horizon agents, capable of complex text synthesis scenarios. However, current evaluation frameworks lack the ability to assess the actual synthesis operations, such as outlining, drafting, and editing. Consequently, they fail to evaluate the actual and detailed capabilities of LLMs. To bridge this gap, we introduce RAVEL, an agentic framework that enables the LLM testers to autonomously plan and execute typical synthesis operations, including outlining, drafting, reviewing, and refining. Complementing this framework, we present C3EBench, a comprehensive benchmark comprising 1,258 samples derived from professional human writings. We utilize a "reverse-engineering" pipeline to isolate specific capabilities across four tasks: Cloze, Edit, Expand, and End-to-End. Through our analysis of 14 LLMs, we uncover that most LLMs struggle with tasks that demand contextual understanding under limited or under-specified instructions. By augmenting RAVEL with SOTA LLMs as operators, we find that such agentic text synthesis is dominated by the LLM's reasoning capability rather than raw generative capacity. Furthermore, we find that a strong reasoner can guide a weaker generator to yield higher-quality results, whereas the inverse does not hold. Our code and data are available at this link: https://github.com/ZhuoerFeng/RAVEL-Reasoning-Agents-Text-Eval.