Newest papers

BetaMem: Learning to Remember a Changing World for Embodied Agents

Mai Liu, He Sun, Siru Zhong, Kerun Chen, Hao Yan, Xiaohui Lv, Shuai Tao, Wulong Liu · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.26554 · cited by 0 works (OpenAlex) · open-access PDF

Portuguese

Os agentes embutidos em contrato de longo prazo da categoria EMBen enfrentam confusão temporal e registros incompletos ou distorcidos, erros recorrentes de regras de memória fixa e custos crescentes de armazenamento e recuperação. Nós apresentamos BetaMem, um quadro para a construção e evolução de memória de longo prazo rastreável através de seis camadas conectadas: trabalho, espaço, entidade, registros episódicos, experiências e memória procedimental. Um contrato de escrita de duplo nível restringe a executabilidade operacional e o suporte de fonte para o conteúdo: o ajustamento fino supervisionado aprende operações de memória compatíveis, enquanto o aprendizado de reforço combina amostragem repetida com a verificação do professor das observações originais para recompensar registros precisos, episódicos e penaliz

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

Abstract

Long-horizon embodied agents face temporal confusion and incomplete or distorted records, recurring errors from fixed memory rules, and growing storage and retrieval costs. We present BetaMem, a framework for building and evolving traceable long-term memory through six linked layers: working, spatial, entity, episodic, experiential, and procedural memory. A dual-level write contract constrains operation executability and source support for content: supervised fine-tuning learns compliant memory operations, while reinforcement learning combines repeated sampling with teacher verification of the original observations to reward complete, accurate records and penalize omissions, distortion, and unsupported content. We formulate failure-driven policy recursive self-improvement through diagnosis along the causal sequence of memory operations and local rule repairs gated by independent replay. Atomic tools frame storage, updates, retrieval, and forgetting as cost-utility allocation under storage and read budgets. Using the open-source Qwen3.5-9B model for memory generation, BetaMem achieves overall scores of 48.08% on M3-Bench-robot, 92.08% on LoCoMo, 72.6% on OpenEQA EM-EQA, and 62.4% on EgoLifeQA, the highest among the listed memory methods under their respective evaluation protocols. Training the model for memory generation raises the contract pass rate from 50.9% to 86.2% while reducing output tokens per stored action by 47.3%. A separate Qwen3-VL-8B-Instruct evaluation on EMem-Bench achieves an average success rate of 38.4% and an average error recurrence rate of 24.9%. We perform best among evaluated methods using comparably sized open-source models. Category results show that we have advantages in integrating information across observations, recalling events over long histories, and inferring personal habits.

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

Topics: Forgetting, Computer science, Reinforcement learning, Artificial intelligence, Embodied cognition, Memory model

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