Newest papers

Where Should a Stuck Agent Spend Its Next Tokens? Causal Attribution of Inference-Time Resources

Saiyue Lyu, Jerrold Huang, Yunbei Zhang, Janet Wang, Zhenting Wang, Yingqiang Ge, Haibo Zhang · 2026 · arXiv

DOI: 10.48550/arXiv.2610.28834 · open-access PDF

German

Inferenzzeit-Scaling wird oft als Token-Budget-Problem behandelt, aber ein fehlgeschlagener Codierungsagent kann qualitativ unterschiedliche Ressourcen suchen: Berechnung, Informationen oder Feedback. Wir führen Banzhaf Resource Attribution (BRA) ein, ein kontrolliertes Faktorrahmen, das den marginalen Beitrag jeder Ressource zur Reparatur und ihre Interaktionen misst und gleichzeitig den fehlgeschlagenen Zustand und das Revisionsprotokoll festhält. Wir bewerten sechs Modelle auf 194 Funktionsstufenfehlern von MBPP+ und HumanEval+ und 126 Repository-Level-Fehlern von SWE-bench Verified.

Machine translation (Meta M2M100); the English original follows.

Abstract

Inference-time scaling is often treated as a token-budget problem, but a failed coding agent can seek qualitatively different resources: computation, information, or feedback. We introduce Banzhaf Resource Attribution (BRA), a controlled factorial framework that measures each resource's marginal contribution to repair and its interactions while holding the failed state and revision protocol fixed. We evaluate six models on 194 function-level failures from MBPP+ and HumanEval+ and 126 repository-level failures from SWE-bench Verified. Task-relevant specifications consistently improve repair, wh

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Is this paper real? This page was built from OpenAlex and arXiv records, not generated. Software can verify any reference in one call: GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.28834, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txt

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange