Tag
MAVEN is a lightweight symbolic reasoning scaffold that improves generalization in agentic tool calling by using modular verification and adaptive tool orchestration. It achieves significant accuracy gains on a new stress-test benchmark (MAVEN-Bench) and remains competitive with proprietary models at a fraction of the cost.
This paper proposes a symbolic framework that converts redacted police narratives into evidence-linked facts using ontology, semantic parsing (AMR), and reasoning, enabling structured querying of incident details that are typically only available in free text.
Proposes BISON, a system combining learned low-level neural policies with high-level symbolic planning for long-horizon embodied tasks, showing strong generalization and efficiency.
This paper introduces the Neural Rule Inducer (NRI), a foundation model for zero-shot logical rule induction that uses domain-agnostic statistical properties to generalize across tasks without retraining.