Newest papers

Self-Credit: Credit Assignment for Long-Horizon Agentic RL via Intra-group Rescaling

Fangcong Yin, Ming Yin, Tanvi Bhandarkar, Aashu Singh, Eryk Helenowski, Shlok Kumar Mishra, Saghar Hosseini, Karthik Abinav Sankararaman · 2026 · arXiv

DOI: 10.48550/arXiv.2610.28388 · open-access PDF

Portuguese

Os agentes do LLM são geralmente treinados com o RLVR, mas a natureza escassa das recompensas típicas do RLVR torna a atribuição de crédito desafiadora, especialmente para tarefas de longo prazo. Propomos Self-Credit, um método de agente RL end-to-end que cria sinais de supervisão densos a partir de recompensas de resultados escassos usando atribuição de crédito auto-gerada através de redefinição intra-grupo. O Self-Credit baseia-se em RL baseado em grupo com três componentes. Primeiro, durante seus lançamentos, a política do modelo gera resumos com citações a turnos anteriores contendo evidências críticas, que servem como crédito auto-gerado. Segundo, refinamos este mo

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

Abstract

LLM agents are commonly trained with RLVR, but the sparse nature of typical RLVR rewards makes credit assignment challenging, especially for long-horizon tasks. We propose Self-Credit, an end-to-end agentic RL method that creates dense supervision signals from sparse outcome rewards using self-generated credit assignment via intra-group rescaling. Self-Credit builds on group-based RL with three components. First, during its rollouts, the policy model generates summaries with citations to earlier turns containing critical evidence, which serve as self-generated credit. Second, we refine this mo

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=2610.28388, 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