Este conteúdo apresenta CGD-PD, uma camada leve para modelos de linguagem grandes (LLMs) que melhora a resposta a perguntas lógicas de três vias (Verdadeiro/Falso/Desconhecido). Ele aborda falhas recorrentes como inconsistência de negação e previsões 'Desconhecido' epistêmicas, utilizando decisões consistentes e desambiguação baseada em prova para maior precisão.
This paper proposes a method for anomaly detection called Chimera Training, focusing on violations of semantic constraints given as logical rules over learned visual concepts. It employs a neural rule evaluator that compiles constraints into directed acyclic graphs, learning logical operators to calculate rule-satisfaction probabilities, even with scarce training data for actual violations.
This research introduces a symbolic reasoning scaffold to address systematic limitations in LLMs' structured logical reasoning, such as conflating hypothesis generation and propagating weak inferences. It operationalizes Peirce's tripartite inference, enforcing logical consistency through algebraic invariants, notably the 'Weakest Link bound' to prevent conclusion reliability from exceeding its least-supported premise.
This study investigates how both AI usage and informativeness shape skill development in logical reasoning tasks. It finds that greater AI usage correlates with weaker skill development, particularly with low-information AI, while high-information AI can improve short-term performance without negatively impacting post-AI outcomes on average.
Este conteúdo aborda um estudo sobre o sistema DeltaLogic, que investiga como pequenas alterações em premissas revelam falhas na revisão de crenças em modelos de raciocínio lógico de IA.