# DiagFlowBench: Evaluating How Language Models Handle Off-Procedure Inputs in Grounded Diagnostic Dialogue

arXiv: https://arxiv.org/abs/2606.17904 · HTML: https://mlsanity.com/paper/2606.17904/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 22 of its 53 references that are in the map.
- First posted on arXiv 16 Jun 2026.

## Builds on (most cited first, 22 of 22)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities](https://mlsanity.com/paper/2507.06261/) (2025)
- [G-Eval: NLG Evaluation using GPT-4 with Better Human Alignment](https://mlsanity.com/paper/2303.16634/) (2023)
- [Know What You Don't Know: Unanswerable Questions for SQuAD](https://mlsanity.com/paper/1806.03822/) (2018)
- [gpt-oss-120b & gpt-oss-20b Model Card](https://mlsanity.com/paper/2508.10925/) (2025)
- [MultiWOZ -- A Large-Scale Multi-Domain Wizard-of-Oz Dataset for Task-Oriented Dialogue Modelling](https://mlsanity.com/paper/1810.00278/) (2018)
- [Retrieval Augmentation Reduces Hallucination in Conversation](https://mlsanity.com/paper/2104.07567/) (2021)
- [LLM Evaluators Recognize and Favor Their Own Generations](https://mlsanity.com/paper/2404.13076/) (2024)
- [LLMs Get Lost In Multi-Turn Conversation](https://mlsanity.com/paper/2505.06120/) (2025)
- [An Evaluation Dataset for Intent Classification and Out-of-Scope Prediction](https://mlsanity.com/paper/1909.02027/) (2019)
- [The Power of Noise: Redefining Retrieval for RAG Systems](https://mlsanity.com/paper/2401.14887/) (2024)
- [Don't Hallucinate, Abstain: Identifying LLM Knowledge Gaps via Multi-LLM Collaboration](https://mlsanity.com/paper/2402.00367/) (2024)
- [AbstentionBench: Reasoning LLMs Fail on Unanswerable Questions](https://mlsanity.com/paper/2506.09038/) (2025)
- [Question Answering for Privacy Policies: Combining Computational and Legal Perspectives](https://mlsanity.com/paper/1911.00841/) (2019)
- [Robustness Testing of Language Understanding in Task-Oriented Dialog](https://mlsanity.com/paper/2012.15262/) (2020)
- [End-to-End Learning of Flowchart Grounded Task-Oriented Dialogs](https://mlsanity.com/paper/2109.07263/) (2021)
- [FlowAgent: Achieving Compliance and Flexibility for Workflow Agents](https://mlsanity.com/paper/2502.14345/) (2025)
- [GuideBench: Benchmarking Domain-Oriented Guideline Following for LLM Agents](https://mlsanity.com/paper/2505.11368/) (2025)
- [CGoDial: A Large-Scale Benchmark for Chinese Goal-oriented Dialog Evaluation](https://mlsanity.com/paper/2211.11617/) (2022)
- [Chitchat as Interference: Adding User Backstories to Task-Oriented Dialogues](https://mlsanity.com/paper/2402.15248/) (2024)
- [PFDial: A Structured Dialogue Instruction Fine-tuning Method Based on UML Flowcharts](https://mlsanity.com/paper/2503.06706/) (2025)

## Abstract (arXiv)

Language models increasingly serve as advisory systems in maintenance operations. To prevent hallucination, established approaches ground these models in procedural documentation, constraining them to prescribed sequences. In practice, however, operators may stray from these steps, requiring models to recognise unscripted, off-procedure utterances. Current benchmarks rarely prioritise this capability. We introduce DiagFlowBench, a dataset of 50 industrial diagnostic flowcharts from a consumer manufacturer converted into 1,676 multi-turn conversations that contrast compliant with off-procedure utterances. Evaluating a panel of ten commercial and open-weight models reveals high variability in abstention rates, with models often selecting a real but contextually inadequate step rather than fabricating facts. The inherent plausibility and authority of this mapped but wrong advice exposes a challenging vulnerability for grounding systems. Code and data are released at https://github.com/guille-gil/DiagFlowBench.