Agent Step Value: Auditing Evaluator-Channel Reversals in Black-Box Agent Traces

• URL: http://arxiv.org/abs/2607.04419v3
• Date: Mon, 13 Jul 2026 16:54:28 GMT
• Title: Agent Step Value: Auditing Evaluator-Channel Reversals in Black-Box Agent Traces
• Abstract summary: When evaluator-derived step rewards are pooled or compared across scoring channels, their sign is treated as transportable.<n>We define Agent Step Value (ASV) as channel-indexed target-margin gain and identify the missing state-by-channel interaction.
• Score: 2.4293637634506404
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: When evaluator-derived step rewards are pooled or compared across scoring channels, their sign is treated as transportable. Yet the same frozen transition can change sign with the scoring channel. Process rewards vary agent states, while evaluator audits vary scoring configurations; neither first difference isolates their interaction. We define Agent Step Value (ASV) as channel-indexed target-margin gain and identify the missing state-by-channel interaction by replaying complete faces. Of 1,100 PubMed open question-answering transitions, 1,004 were complete across four cyclic layouts. Their mean update is positive under direct scoring (+0.163 [0.102, 0.218]) and negative with an externally generated view (-0.160 [-0.244, -0.079]), giving an interaction of -0.323 [-0.418, -0.232]. Across paired transitions, 507/1,004 cross zero in one direction or the other. Matched templates isolate a generated-minus-quote interaction of -1.121 [-1.534, -0.703]. Its direction remains negative after the readout and evaluator-stack changes, and quote yields a higher paired area under the receiver operating characteristic curve (AUC) for stored success at all three bridge vertices. Because the two views preserve the same task information conditional on the retained state, a representation-invariance null predicts equal responses and success rankings across them. ASV provides a transport audit for evaluator-derived step measurements; causal actor credit lies outside its estimand.

Related papers

• Architecture-Dependent Causal Transfer of Activation States Across Large Language Models [51.56484100374058]
  Internal language activation states can be transferred causally between different large language models.<n>Injecting projected activations into a target model during generation produces a statistically significant, preregistered causal effect.<n>We conclude that end-to-end activation-state transfer between LLMs, as currently implemented, is architecture-dependent rather than universal.
  arXiv  Detail & Related papers  (2026-08-17T09:53:27Z)
• A Pre-Specified Construction-Confirmation Test of Operation-Level Causal Transfer Across Finite Isomorphic Symbolic Domains [0.0]
  We ask if the hidden-state difference between two operations is estimated separately for each source input, does adding that difference move the model toward the corresponding recipient answer?<n>The design compares this input-specific intervention with wrong-operation, norm-matched random, and no-op controls, and separates candidate construction from an independently isolated confirmation split.
  arXiv  Detail & Related papers  (2026-08-16T15:37:26Z)
• Stage-Replay Divergence Follows the KV Cache: Fixed-Prefix Precision Controls and Bidirectional Cache Transplantation [51.56484100374058]
  Stage-replay reconstructs intermediate token prefixes and treat fresh-prefill continuation as continuation from the decoder state that originally reached the prefix.<n>A matched 200-item experiment compares retained live cache with one-shot prefill of identical integer tokens and places an exact replica on both sides.
  arXiv  Detail & Related papers  (2026-07-30T16:41:40Z)
• When Do Agent Loops Mistake Stagnation for Progress? Self-Evaluation Bias and Externally Grounded Verification in Long-Running Autonomous LLM Agent Loops [0.0]
  Long-running autonomous agents plan, act, and judge their own completion without human intervention.<n>When an agent grades its own work, self-evaluation bias takes hold: plausible changes are accepted as progress while real-world outcomes stagnate or regress.<n>We build a testbed that holds the agent and its tool surface fixed and manipulates only the information-channel type of the evaluator that gates the loop.
  arXiv  Detail & Related papers  (2026-07-27T23:52:15Z)
• ScratchWorld: Evaluating If World Models Compute Executable Consequences [2.3981743514438087]
  We introduce ScratchWorld, an offline diagnostic benchmark that treats Scratch projects as executable worlds.<n> ScratchWorld evaluates next-state prediction, long-horizon tracking, causal event attribution, and counterfactual prediction.
  arXiv  Detail & Related papers  (2026-06-30T14:02:25Z)
• Between Zeros and Ones: Behavioral Characterization Beyond Binary Labeling Across Public ICS Datasets [0.12249546377051435]
  Intrusion detection in Industrial Control Systems is typically evaluated on a small set of public benchmarks using binary normal'' versus attack'' labels.<n>We propose a behavioral characterization framework that maps raw multivariate process traces into five interpretable physical primitives.<n>We show that attack windows exhibit clear behavioral shifts relative to normal operation while the three datasets occupy largely distinct regions of the behavioral space.
  arXiv  Detail & Related papers  (2026-06-29T15:58:34Z)
• Efficient Visual Pointing for Embodied AI:Agent-Driven Data Synthesis, Cross-Block Attention, and Iterative Correction [55.11480729304395]
  We describe our PointArena 2026 solution, which achieves 77.2% overall accuracy and ranks second on the benchmark.<n>The ap proach targets three failure modes. First, agent-driven syn thesis builds large semantic and anchor-relative candidate pools.<n>Second, a determinis tic steerable-data pipeline creates a verified 10,000-sample main set, plus reserve samples, using masks, templates, and path verification.
  arXiv  Detail & Related papers  (2026-06-29T06:39:03Z)
• Online Reward-Punishment Learning from Fixed-Channel Perceptual Event Streams without Environment Rewards [3.1050763275397806]
  OHIRL separates four roles: M_psi learns next-packet prediction, D_omega models residual dynamics, C_eta is a fixed internal post-transition trajectory.<n>C_eta uses a recovery-positive and persistence/growth-negative residual-regulation orientation.<n>A conditional error decomposition separates B_xi evidence-estimation error from residual policy-optimization error.
  arXiv  Detail & Related papers  (2026-06-17T11:43:10Z)
• Multimodal Evaluator Preference Collapse: Cross-Modal Contagion in Self-Evolving Agents [1.0152838128195467]
  We show that Evaluator Preference Collapse ( EPC) is dramatically amplified in multimodal settings.<n>We then demonstrate a novel phenomenon we term cross-modal contagion.<n>We identify cross-model evaluator architecture as the primary risk factor for preference drift.
  arXiv  Detail & Related papers  (2026-06-15T13:18:20Z)
• StainFlow: Entity-Stain Tracking and Evidence Linking for Process Rewards in GUI Agents [67.03593791535786]
  Reinforcement Learning (RL) has become a promising approach for improving GUI Agents in long-horizon, digital environments.<n>To mitigate this issue, recent studies introduce Process Reward Models (PRMs)<n>PRMs provide finer-grained training feedback through global milestone verification or local step-level evaluation.<n>We propose StainFlow, an entity-stain-flow process reward model for GUI Agents.
  arXiv  Detail & Related papers  (2026-06-05T08:17:28Z)
• What to Test Next: Interpretable Coverage Gap Discovery in Driving VLMs [52.50210189669399]
  Driving vision-language models (VLMs) must accurately understand scenes across diverse conditions defined by Operational Design Domains (ODDs)<n>We propose SliceScorer, a deterministic scoring rule for missing-slice recommendation.<n>We show that SliceNav surfaces high-risk coverage gaps more effectively than prior slice-discovery methods.
  arXiv  Detail & Related papers  (2026-06-01T03:18:01Z)
• TraceGraph: Shared Decision Landscapes for Diagnosing and Improving Agent Trajectories [61.51984029109709]
  We introduce TraceGraph, a graph-based framework that turns released multi-model agent trajectories into shared decision landscapes.<n>For each task, TraceGraph builds a graph over observable action-observation states from pooled rollouts before model identity is introduced.<n>It then overlays outcome-informed productive cores and trap regions, and summarizes each rollout with three events: Access, Trap exposure, and Repair.
  arXiv  Detail & Related papers  (2026-05-29T13:40:31Z)
• ChromaFlow: A Negative Ablation Study of Orchestration Overhead in Tool-Augmented Agent Evaluation [0.0]
  This report presents a tool-augmented autonomous reasoning framework built around planner-directed execution, specialized tool use, and telemetry-driven evaluation.<n>We analyze ChromaFlow on GAIA 2023 Level-1 validation tasks under clean evaluation constraints.
  arXiv  Detail & Related papers  (2026-05-13T20:40:37Z)
• Beyond Completion: Probing Cumulative State Tracking to Predict LLM Agent Performance [9.771590610969918]
  We introduce Working Memory Fidelity-Active Manipulation (WMF-AM)<n>We evaluate it on 20 open-weight models (0.5B-35B, 13 families) against a released deterministic 10-task agent battery.
  arXiv  Detail & Related papers  (2026-03-28T17:25:11Z)
• TRAJEVAL: Decomposing Code Agent Trajectories for Fine-Grained Diagnosis [23.834704102474927]
  Code agents can resolve GitHub issues, yet when they fail, current evaluation provides no visibility into where or why.<n>We introduce TRAJEVAL, a diagnostic framework that decomposes agent trajectories into three interpretable stages.<n>We validate that these diagnostics are predictive, achieving model-level Pass@1 prediction within 0.87-2.1% MAE.
  arXiv  Detail & Related papers  (2026-03-25T05:27:03Z)
• CARE What Fails: Contrastive Anchored-REflection for Verifiable Multimodal [84.71254539482369]
  Group-relative reinforcement learning with verifiable rewards (RLVR) often wastes the most informative data it already has the failures.<n>We present CARE, a failure-centric post-training framework for multimodal reasoning that turns errors into supervision.<n> CARE improves accuracy and training smoothness while explicitly increasing the share of learning signal that comes from failures.
  arXiv  Detail & Related papers  (2025-12-22T16:34:21Z)
• Prune-Then-Plan: Step-Level Calibration for Stable Frontier Exploration in Embodied Question Answering [52.69447404069251]
  Large vision-language models (VLMs) have improved embodied question answering (EQA) agents by providing strong semantic priors for open-vocabulary reasoning.<n>We propose Prune-Then-Plan, a framework that stabilizes exploration through step-level calibration.
  arXiv  Detail & Related papers  (2025-11-24T22:50:50Z)
• Rethinking Metrics and Benchmarks of Video Anomaly Detection [58.37571339811799]
  Video Anomaly Detection (VAD) aims to detect anomalies that deviate from expectation.<n>Existing VAD metrics are influenced by single annotation bias.<n>Existing benchmarks lack the capability to evaluate scene overfitting of fully/weakly-supervised algorithms.
  arXiv  Detail & Related papers  (2025-05-25T08:09:42Z)
• Constructing interval variables via faceted Rasch measurement and multitask deep learning: a hate speech application [63.10266319378212]
  We propose a method for measuring complex variables on a continuous, interval spectrum by combining supervised deep learning with the Constructing Measures approach to faceted Rasch item response theory (IRT) We demonstrate this new method on a dataset of 50,000 social media comments sourced from YouTube, Twitter, and Reddit and labeled by 11,000 U.S.-based Amazon Mechanical Turk workers.
  arXiv  Detail & Related papers  (2020-09-22T02:15:05Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.