# CoDial: Interpretable Task-Oriented Dialogue Systems Through Dialogue Flow Alignment

arXiv: https://arxiv.org/abs/2506.02264 · HTML: https://mlsanity.com/paper/2506.02264/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in LLM preference optimization methods, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 20 of its 40 references that are in the map.
- First posted on arXiv 2 Jun 2025.

## Placed in

- LLM alignment / RLHF / preference optimisation / [LLM preference optimization methods](https://mlsanity.com/alignment/llm-preference-optimization-methods/)
- LLM safety and security / [Safety alignment, refusal and guardrails](https://mlsanity.com/safety/safety-alignment-refusal-and-guardrails/)

## Builds on (most cited first, 20 of 20)

- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [A Call for Clarity in Reporting BLEU Scores](https://mlsanity.com/paper/1804.08771/) (2018)
- [Is Your Code Generated by ChatGPT Really Correct? Rigorous Evaluation of Large Language Models for Code Generation](https://mlsanity.com/paper/2305.01210/) (2023)
- [MultiWOZ -- A Large-Scale Multi-Domain Wizard-of-Oz Dataset for Task-Oriented Dialogue Modelling](https://mlsanity.com/paper/1810.00278/) (2018)
- [NeMo Guardrails: A Toolkit for Controllable and Safe LLM Applications with Programmable Rails](https://mlsanity.com/paper/2310.10501/) (2023)
- [A Survey on Dialogue Systems: Recent Advances and New Frontiers](https://mlsanity.com/paper/1711.01731/) (2017)
- [Opportunities and Challenges for ChatGPT and Large Language Models in Biomedicine and Health](https://mlsanity.com/paper/2306.10070/) (2023)
- [Are LLMs All You Need for Task-Oriented Dialogue?](https://mlsanity.com/paper/2304.06556/) (2023)
- [Shades of BLEU, Flavours of Success: The Case of MultiWOZ](https://mlsanity.com/paper/2106.05555/) (2021)
- [STAR: A Schema-Guided Dialog Dataset for Transfer Learning](https://mlsanity.com/paper/2010.11853/) (2020)
- [SGP-TOD: Building Task Bots Effortlessly via Schema-Guided LLM Prompting](https://mlsanity.com/paper/2305.09067/) (2023)
- [Large Language Models as Zero-shot Dialogue State Tracker through Function Calling](https://mlsanity.com/paper/2402.10466/) (2024)
- [End-to-end Task-oriented Dialogue: A Survey of Tasks, Methods, and Future Directions](https://mlsanity.com/paper/2311.09008/) (2023)
- [Schema-Guided Paradigm for Zero-Shot Dialog](https://mlsanity.com/paper/2106.07056/) (2021)
- [Workflow Discovery from Dialogues in the Low Data Regime](https://mlsanity.com/paper/2205.11690/) (2022)
- [Mars: Modeling Context & State Representations with Contrastive Learning for End-to-End Task-Oriented Dialog](https://mlsanity.com/paper/2210.08917/) (2022)
- [AnyTOD: A Programmable Task-Oriented Dialog System](https://mlsanity.com/paper/2212.09939/) (2022)
- [DARD: A Multi-Agent Approach for Task-Oriented Dialog Systems](https://mlsanity.com/paper/2411.00427/) (2024)
- [Unsupervised Learning of Hierarchical Conversation Structure](https://mlsanity.com/paper/2205.12244/) (2022)
- [Unsupervised Extraction of Dialogue Policies from Conversations](https://mlsanity.com/paper/2406.15214/) (2024)

## Built on by (most cited first, 1 of 1)

- [RedDebate: Safer Responses Through Multi-Agent Red Teaming Debates](https://mlsanity.com/paper/2506.11083/) (2025)

## Abstract (arXiv)

Building Task-Oriented Dialogue (TOD) systems that generalize across different tasks remains a challenging problem. Data-driven approaches often struggle to transfer effectively to unseen tasks. While recent schema-based TOD frameworks improve generalization by decoupling task logic from language understanding, their reliance on neural or generative models often obscures how task schemas influence behaviour and hence impair interpretability. In this work, we introduce a novel framework, CoDial (Code for Dialogue), at the core of which is converting a predefined task schema to a structured heterogeneous graph and then to programmatic LLM guardrailing code, such as NVIDIA's Colang. The pipeline enables efficient and interpretable alignment of dialogue policies during inference. We introduce two paradigms for LLM guardrailing code generation, CoDialfree and CoDialstructured, and propose a mechanism that integrates human feedback to iteratively improve the generated code. Empirically, CoDial achieves state-of-the-art (SOTA) performance on the widely used benchmark datasets, while providing inherent interpretability in the design. We additionally demonstrate CoDial's iterative improvement via manual and LLM-aided feedback, making it a practical tool for human-guided alignment of LLMs in unseen domains.