# When Evidence is Sparse: Weakly Supervised Early Failure Alerting in Dialogs and LLM-Agent Trajectories

arXiv: https://arxiv.org/abs/2606.05414 · HTML: https://mlsanity.com/paper/2606.05414/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in Software engineering coding agents, one of 16 directions in LLM agents and tool use.
- Builds on 19 of its 45 references that are in the map.
- First posted on arXiv 3 Jun 2026.

## Placed in

- LLM agents and tool use / [Software engineering coding agents](https://mlsanity.com/agents/software-engineering-coding-agents/)

## Builds on (most cited first, 19 of 19)

- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [Let's Verify Step by Step](https://mlsanity.com/paper/2305.20050/) (2023)
- [A Survey on Large Language Model based Autonomous Agents](https://mlsanity.com/paper/2308.11432/) (2023)
- [ALFWorld: Aligning Text and Embodied Environments for Interactive Learning](https://mlsanity.com/paper/2010.03768/) (2020)
- [Attention-based Deep Multiple Instance Learning](https://mlsanity.com/paper/1802.04712/) (2018)
- [Math-Shepherd: Verify and Reinforce LLMs Step-by-step without Human Annotations](https://mlsanity.com/paper/2312.08935/) (2023)
- [AppWorld: A Controllable World of Apps and People for Benchmarking Interactive Coding Agents](https://mlsanity.com/paper/2407.18901/) (2024)
- [Judging the Judges: A Systematic Study of Position Bias in LLM-as-a-Judge](https://mlsanity.com/paper/2406.07791/) (2024)
- [Persuasion for Good: Towards a Personalized Persuasive Dialogue System for Social Good](https://mlsanity.com/paper/1906.06725/) (2019)
- [pymoo: Multi-objective Optimization in Python](https://mlsanity.com/paper/2002.04504/) (2020)
- [Multiple Instance Learning: A Survey of Problem Characteristics and Applications](https://mlsanity.com/paper/1612.03365/) (2016)
- [Reward-Conditioned Policies](https://mlsanity.com/paper/1912.13465/) (2019)
- [Provable Benefit of Orthogonal Initialization in Optimizing Deep Linear Networks](https://mlsanity.com/paper/2001.05992/) (2020)
- [Joint Turn and Dialogue level User Satisfaction Estimation on Multi-Domain Conversations](https://mlsanity.com/paper/2010.02495/) (2020)
- [Stop&Hop: Early Classification of Irregular Time Series](https://mlsanity.com/paper/2208.09795/) (2022)
- [Learning the Optimal Stopping for Early Classification within Finite Horizons via Sequential Probability Ratio Test](https://mlsanity.com/paper/2501.18059/) (2025)
- [Multi-Step Dialogue Workflow Action Prediction](https://mlsanity.com/paper/2311.09593/) (2023)

## Built on by (most cited first, 4 of 4)

- [Real-Time Detection and Repair of LLM Agent Failures](https://mlsanity.com/paper/2608.02464/) (2026)
- [Fail-Fast, Restart-Smart: Early Failure Prediction and Restart for SWE Agentic Tasks](https://mlsanity.com/paper/2608.03222/) (2026)
- [Doomed from the Start: Early Abort of LLM Agent Episodes via a Recall-Controlled Probe Cascade](https://mlsanity.com/paper/2607.06503/) (2026)
- [Dream2Discern: Learning from Hard Latent Variants for Runtime Failure Prediction in Multi-Agent Systems](https://mlsanity.com/paper/2610.16071/) (2026)

## Abstract (arXiv)

Early failure alerting requires deciding, while a dialog or agent trajectory is still unfolding, whether to flag it as likely to fail. This is challenging because supervision is typically available only as a trajectory-level success/failure label while alerts must be raised from partial interactions. Prior early-classification methods often bridge this gap by assigning the terminal label to every prefix, treating every turn as failure evidence. We hypothesize that this prefix-label assumption is poorly matched to multi-turn language interactions, where evidence of eventual failure is sparse and often delayed. In this paper, we introduce a two-stage approach that learns from this sparse evidence structure and uses the resulting risk estimates for controllable early alerting. Specifically, our attention-based failure predictor learns sparse turn-level failure evidence from trajectory labels and uses it to estimate failure risk from partial histories. We then pair this predictor with α-STOP, a single preference-conditioned stopping policy that selects an accuracy-earliness operating point at inference time rather than training a separate trigger for each preference. Across five benchmarks spanning customer support, task-oriented dialog, persuasion, tool use, and planning, we first show that high-relevance failure evidence occupies only 4.7-11.3% of turns and first appears after 59.0-83.6% of trajectories on average. We further show that the attention-based predictor improves Pareto-frontier quality (hypervolume) by 1-10% over naive prefix supervision, and that the full system improves frontier quality by 3-42% over state-of-the-art trigger policies while reducing training cost per operating point by 1-3 orders of magnitude.