# Localizing Prompt Ambiguity in Large Language Models with Probe-Targeted Attribution

arXiv: https://arxiv.org/abs/2606.05486 · HTML: https://mlsanity.com/paper/2606.05486/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 33 of its 45 references that are in the map.
- First posted on arXiv 3 Jun 2026.

## Builds on (most cited first, 30 of 33)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [A Unified Approach to Interpreting Model Predictions](https://mlsanity.com/paper/1705.07874/) (2017)
- [Measuring Mathematical Problem Solving With the MATH Dataset](https://mlsanity.com/paper/2103.03874/) (2021)
- ["Why Should I Trust You?": Explaining the Predictions of Any Classifier](https://mlsanity.com/paper/1602.04938/) (2016)
- [Axiomatic Attribution for Deep Networks](https://mlsanity.com/paper/1703.01365/) (2017)
- [Learning Important Features Through Propagating Activation Differences](https://mlsanity.com/paper/1704.02685/) (2017)
- [Understanding intermediate layers using linear classifier probes](https://mlsanity.com/paper/1610.01644/) (2016)
- [Attention is not Explanation](https://mlsanity.com/paper/1902.10186/) (2019)
- [Quantifying Attention Flow in Transformers](https://mlsanity.com/paper/2005.00928/) (2020)
- [AmbigQA: Answering Ambiguous Open-domain Questions](https://mlsanity.com/paper/2004.10645/) (2020)
- [Layer by Layer: Uncovering Hidden Representations in Language Models](https://mlsanity.com/paper/2502.02013/) (2025)
- [SWE-Bench Pro: Can AI Agents Solve Long-Horizon Software Engineering Tasks?](https://mlsanity.com/paper/2509.16941/) (2025)
- [ASQA: Factoid Questions Meet Long-Form Answers](https://mlsanity.com/paper/2204.06092/) (2022)
- [AtP*: An efficient and scalable method for localizing LLM behaviour to components](https://mlsanity.com/paper/2403.00745/) (2024)
- [Guided Integrated Gradients: An Adaptive Path Method for Removing Noise](https://mlsanity.com/paper/2106.09788/) (2021)
- [LeetCodeDataset: A Temporal Dataset for Robust Evaluation and Efficient Training of Code LLMs](https://mlsanity.com/paper/2504.14655/) (2025)
- [Selectively Answering Ambiguous Questions](https://mlsanity.com/paper/2305.14613/) (2023)
- [No Answer Needed: Predicting LLM Answer Accuracy from Question-Only Linear Probes](https://mlsanity.com/paper/2509.10625/) (2025)
- [AMBROSIA: A Benchmark for Parsing Ambiguous Questions into Database Queries](https://mlsanity.com/paper/2406.19073/) (2024)
- [What Prompts Don't Say: Understanding and Managing Underspecification in LLM Prompts](https://mlsanity.com/paper/2505.13360/) (2025)
- [ICR Probe: Tracking Hidden State Dynamics for Reliable Hallucination Detection in LLMs](https://mlsanity.com/paper/2507.16488/) (2025)
- [Understanding Integrated Gradients with SmoothTaylor for Deep Neural Network Attribution](https://mlsanity.com/paper/2004.10484/) (2020)
- [TokenShapley: Token Level Context Attribution with Shapley Value](https://mlsanity.com/paper/2507.05261/) (2025)
- [Disambiguate First, Parse Later: Generating Interpretations for Ambiguity Resolution in Semantic Parsing](https://mlsanity.com/paper/2502.18448/) (2025)
- [RelP: Faithful and Efficient Circuit Discovery in Language Models via Relevance Patching](https://mlsanity.com/paper/2508.21258/) (2025)
- [Identifying Unclear Questions in Community Question Answering Websites](https://mlsanity.com/paper/1901.06168/) (2019)
- [Scope Ambiguities in Large Language Models](https://mlsanity.com/paper/2404.04332/) (2024)
- [Probing Syntax in Large Language Models: Successes and Remaining Challenges](https://mlsanity.com/paper/2508.03211/) (2025)
- [Sparse Neurons Carry Strong Signals of Question Ambiguity in LLMs](https://mlsanity.com/paper/2509.13664/) (2025)
- [Generalized Attention Flow: Feature Attribution for Transformer Models via Maximum Flow](https://mlsanity.com/paper/2502.15765/) (2025)

## Abstract (arXiv)

Prompt ambiguity is a common source of failure in large language models, but is difficult to localize because it is a latent property of the prompt, while existing attribution methods are designed to explain observable outputs such as logits or generated tokens. We introduce PRIG, a gradient attribution method that uses a probe logit to attribute latent ambiguity to token positions. Specifically, PRIG trains a linear probe to distinguish clear prompts from ambiguous prompts and attributes the probe score to earlier token representations in the residual stream. To enable token-level evaluation, we construct synthetic ambiguity datasets across coding, math, and writing by rewriting one task-critical sentence per prompt, and complement them with a human-written gold benchmark. In this setting, PRIG localizes ambiguous spans substantially better than gradient attribution baselines, achieving 0.840 AUROC on the combined synthetic benchmark and 0.891 AUROC on the gold set. It also outperforms GPT-5.4 on sentence-level ambiguity identification and retains useful signal out-of-domain. These results establish PRIG as a practical tool for identifying which parts of a prompt are ambiguous. More broadly, they suggest that latent prompt properties can be localized through intermediate representations, rather than through output-level attribution.