Eliminating Agentic Workflow for Introduction Generation with Parametric Stage Tokens

TL;DR

This paper introduces STIG, a novel method that encodes the logical structure of introduction generation directly into LLMs using stage tokens, enabling single-inference multi-stage writing without external workflows.

Contribution

The paper proposes STIG, a parametric approach that replaces external agentic workflows with stage tokens embedded in the model, improving coherence and efficiency in introduction generation.

Findings

STIG outperforms traditional workflows on semantic similarity.

STIG produces more structurally rational introductions.

Single-inference generation reduces complexity and errors.

Abstract

In recent years, using predefined agentic workflows to guide large language models (LLMs) for literature classification and review has become a research focus. However, writing research introductions is more challenging. It requires rigorous logic, coherent structure, and abstract summarization. Existing workflows often suffer from long reasoning chains, error accumulation, and reduced textual coherence. To address these limitations, we propose eliminating external agentic workflows. Instead, we directly parameterize their logical structure into the LLM. This allows the generation of a complete introduction in a single inference. To this end, we introduce the Stage Token for Introduction Generation (STIG). STIG converts the multiple stages of the original workflow into explicit stage signals. These signals guide the model to follow different logical roles and functions during generation. Through instruction tuning, the model learns the mapping between stage tokens and text functions. It also learns the logical order and transition patterns between stages, encoding this knowledge into the model parameters. Experimental results show that STIG can generate multi-stage text in a single inference. It does not require explicit workflow calls. STIG outperforms traditional agentic workflows and other baselines on metrics of semantic similarity and sentence-level structural rationality. The code is provided in the Supplementary Materials.