Disagree to Explore, Agree to Commit: Routing-Guided Test-Time Scaling for Software Agents

• URL: http://arxiv.org/abs/2608.22191v1
• Date: Sun, 23 Aug 2026 03:07:03 GMT
• Title: Disagree to Explore, Agree to Commit: Routing-Guided Test-Time Scaling for Software Agents
• Abstract summary: We study whether native MoE router traces can guide steering and selection without an external judge or selection-time test execution.<n>Our analysis shows that routing provides a robust behavioral role signal; token-granular readouts and decision-matched comparison sets turn it into effective control.<n>We evaluate on SWE-bench Verified using open-weight sparse MoE agents across scales and reasoning-effort settings.
• Score: 60.650808962786364
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Software-engineering agents solve repository-level tasks through long, stochastic tool-use trajectories, and repeated attempts often find fixes missed by one run. Test-time scaling is difficult because patches lack canonical answer forms, while sibling actions from a shared prefix are correlated. We study whether native MoE router traces can guide steering and selection without an external judge or selection-time test execution. Our analysis shows that routing provides a robust behavioral role signal; token-granular readouts and decision-matched comparison sets turn it into effective control. We therefore introduce Risa (Routing-Informed Steering and Arbitration): within trajectories, routing encourages diverse exploration and controlled convergence during patch commitment; across separately sampled trajectories, agreement at informative patch positions selects a final candidate. We evaluate on SWE-bench Verified using open-weight sparse MoE agents across scales and reasoning-effort settings. Risa's routing arbitration raises the macro-average resolved rate from 44.9% under uniform sampling to 48.2% on the gpt-oss family, matching text consensus without answer-string matching, and it transfers to Qwen3.6, where it improves on uniform choice and matches text consensus on the full 500-task benchmark.

Related papers

• Multi-Agent Self-Improving Reinforcement Learning for Video Reasoning [45.361080098317665]
  Video reasoning tasks such as grounded video question answering and temporal grounding require selecting temporal evidence that supports the query.<n>We study whether a frozen verifier can also guide training.<n>Our framework couples a trainable emphGrounder with a frozen emphVerifier: the Grounder samples candidate trajectories and evidence segments.
  arXiv  Detail & Related papers  (2026-08-25T07:46:32Z)
• Task- and Session-Level Model Routing: A Common-Interface Hybrid Evaluation of Four Open-Source Routers Across Four Benchmarks [0.0]
  We evaluate four router implementations across RouterBench, BFCL v4, tau2-bench, and WebArena.<n>We evaluate 290 frozen tasks against a locked matrix of 2,610 candidate outcomes.
  arXiv  Detail & Related papers  (2026-07-28T21:29:03Z)
• Set-shifting Behavioral Test for Harnessed Agents [0.0]
  We borrow set-shifting from cognitive psychology to study how well agents adapt to hidden reliability shifts.<n>We find that agents, by default, settle on a small recurring routine within a few turns of each boundary.<n>We score the set-shifting accuracy for each agent trajectory: the joint probability of routing to the target tool group in every post-shift window.
  arXiv  Detail & Related papers  (2026-07-15T02:49:05Z)
• What Resolve Rate Hides: Trajectory Structure Diagnostics for Coding Agents [14.988334585821939]
  Coding agents are ranked almost entirely by resolve rate.<n>A single pass/fail label says nothing about why a run failed or why an accepted run spent extra steps, time, or tokens.<n>We present TraceProbe, a trajectory-diagnostic framework that recovers what resolve rate hides.
  arXiv  Detail & Related papers  (2026-07-07T12:09:46Z)
• ARBOR: Online Process Rewards via a Reusable Rubric Buffer for Search Agents [48.80766702702854]
  LLM-based search agents are trained predominantly with outcome-only reward, leaving the search process itself unsupervised.<n>This signal degenerates on outcome-homogeneous groups where all sampled trajectories share the same correctness, yielding zero within-group advantage and no gradient.<n>We propose ARBOR (Adaptive Buffer for Online Reward), a reusable process-reward framework that maintains a rubric memory shared across queries.
  arXiv  Detail & Related papers  (2026-06-02T06:58:54Z)
• ExComm: Exploration-Stage Communication for Error-Resilient Agentic Test-Time Scaling [57.42714978834704]
  ExComm is a communication protocol for exploration-stage agentic test-time scaling.<n>We show that ExComm consistently outperforms strong test-time scaling baselines.
  arXiv  Detail & Related papers  (2026-05-21T07:38:44Z)
• Scaling Test-Time Compute for Agentic Coding [126.72747643609274]
  We propose a test-time scaling framework for agentic coding based on compact representations of rollout trajectories.<n>Our framework converts each rollout into a structured summary that preserves its salient hypotheses, progress, and failure modes.<n>Our method consistently improves the performance of frontier coding agents across SWE-Bench Verified and Terminal-Bench v2.0.
  arXiv  Detail & Related papers  (2026-04-16T17:39:33Z)
• ORCH: many analyses, one merge-a deterministic multi-agent orchestrator for discrete-choice reasoning with EMA-guided routing [0.6445605125467574]
  ORCH is a framework for discrete-choice reasoning that orchestrates heterogeneous language models.<n>It uses fixed rules for task decomposition and answer aggregation, keeping the pipeline predictable, reproducible, and training-free.<n>Experiments on MMLU, MMLU-Pro, and GSM8K show that ORCH consistently outperforms single-model baselines and a majority-vote ensemble.
  arXiv  Detail & Related papers  (2026-02-02T08:27:58Z)
• What If We Allocate Test-Time Compute Adaptively? [2.1713977971908944]
  Test-time scaling allocates inference computation uniformly, uses fixed sampling strategies, and applies verification only for reranking.<n>We propose a verifier-guided adaptive framework treating reasoning as iterative trajectory generation and selection.<n>Across datasets, our dynamic, PRM-guided approach consistently outperforms direct test-time scaling.
  arXiv  Detail & Related papers  (2026-02-01T07:30:22Z)
• Semi-DETR: Semi-Supervised Object Detection with Detection Transformers [105.45018934087076]
  We analyze the DETR-based framework on semi-supervised object detection (SSOD) We present Semi-DETR, the first transformer-based end-to-end semi-supervised object detector. Our method outperforms all state-of-the-art methods by clear margins.
  arXiv  Detail & Related papers  (2023-07-16T16:32:14Z)
• Detection Transformer with Stable Matching [48.963171068785435]
  We show that the most important design is to use and only use positional metrics to supervise classification scores of positive examples. Under the principle, we propose two simple yet effective modifications by integrating positional metrics to DETR's classification loss and matching cost. We achieve 50.4 and 51.5 AP on the COCO detection benchmark using ResNet-50 backbones under 12 epochs and 24 epochs training settings.
  arXiv  Detail & Related papers  (2023-04-10T17:55:37Z)

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.