Decoupling Task-Solving and Output Formatting in LLM Generation

2025-10-03

AI Summary

Large language models (LLMs) often exhibit task deviation and format violations when executing complex instructions that simultaneously demand logical reasoning and strict syntactic adherence, due to inherent conflicts between task objectives and formatting constraints. To address this, we propose Deco-G—a novel framework that decouples task solving from format compliance: the LLM focuses exclusively on reasoning, while a tractable probabilistic model—built upon an enhanced Trie structure and a pruned Hidden Markov Model—explicitly encodes and enforces output syntax. Instruction-aware knowledge distillation and decoding-time probability fusion enable efficient coordination between the two components. Evaluated on mathematical reasoning, LLM-as-a-judge, and event argument extraction, Deco-G achieves 1.0–6.0% relative performance gains while maintaining 100% format compliance, substantially mitigating the instruction-interference-induced decline in instruction following fidelity.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information

Abstract

Large language models (LLMs) are increasingly adept at following instructions containing task descriptions to solve complex problems, such as mathematical reasoning and automatic evaluation (LLM-as-a-Judge). However, as prompts grow more complex, models often struggle to adhere to all instructions. This difficulty is especially common when instructive prompts intertwine reasoning directives -- specifying what the model should solve -- with rigid formatting requirements that dictate how the solution must be presented. The entanglement creates competing goals for the model, suggesting that more explicit separation of these two aspects could lead to improved performance. To this front, we introduce Deco-G, a decoding framework that explicitly decouples format adherence from task solving. Deco-G handles format compliance with a separate tractable probabilistic model (TPM), while prompts LLMs with only task instructions. At each decoding step, Deco-G combines next token probabilities from the LLM with the TPM calculated format compliance likelihood to form the output probability. To make this approach both practical and scalable for modern instruction-tuned LLMs, we introduce three key innovations: instruction-aware distillation, a flexible trie-building algorithm, and HMM state pruning for computational efficiency. We demonstrate the effectiveness of Deco-G across a wide range of tasks with diverse format requirements, including mathematical reasoning, LLM-as-a-judge, and event argument extraction. Overall, our approach yields 1.0% to 6.0% relative gain over regular prompting practice with guaranteed format compliance.

Problem

• Decoupling task-solving from formatting requirements in LLM generation
• Addressing performance issues from entangled reasoning and formatting instructions
• Ensuring guaranteed format compliance while maintaining task-solving accuracy

Innovation

• Decoupling task-solving and formatting with separate models
• Using tractable probabilistic model for format compliance
• Combining LLM probabilities with format likelihood per step