论文标题
行为克隆的变压器是神经符号推理器
Behavior Cloned Transformers are Neurosymbolic Reasoners
论文作者
论文摘要
在这项工作中，我们探索了通过符号模块的信息增强交互式代理的技术，就像人类使用计算器和GPS系统之类的工具一样，可以协助算术和导航。我们在文本游戏中测试了代理商的能力 - 挑战基准，以评估基于语言的环境中游戏代理的多步推理能力。我们的实验研究表明，将这些符号模块的动作注射到行为克隆的变压器代理的动作空间中，可以提高四个文本游戏基准的性能，以测试算术，导航，分类，分类和常识推理平均22％，从而使代理在未观察的游戏中达到最高可能的性能。该动作注入技术很容易扩展到新的代理，环境和符号模块。
In this work, we explore techniques for augmenting interactive agents with information from symbolic modules, much like humans use tools like calculators and GPS systems to assist with arithmetic and navigation. We test our agent's abilities in text games -- challenging benchmarks for evaluating the multi-step reasoning abilities of game agents in grounded, language-based environments. Our experimental study indicates that injecting the actions from these symbolic modules into the action space of a behavior cloned transformer agent increases performance on four text game benchmarks that test arithmetic, navigation, sorting, and common sense reasoning by an average of 22%, allowing an agent to reach the highest possible performance on unseen games. This action injection technique is easily extended to new agents, environments, and symbolic modules.
