Behavior Cloned Transformers are Neurosymbolic Reasoners
By Ruoyao Wang, Peter Alexander Jansen, Marc-Alexandre Côté, Prithviraj Ammanabrolu. 2022
14 citations recorded by the source corpus.
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.
Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Natural Language Processing, Knowledge Graphs & Representation
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Mohit Bansal
- Timo Schick
- Luke S. Zettlemoyer
- Hinrich Schütze
- Thomas Scialom
- Yejin Choi
- Asli Celikyilmaz
- Jianfeng Gao
In Computer Vision and Pattern Recognition
- Ramakanth Pasunuru
- Mohit Bansal
- Asli Celikyilmaz
- Edouard Grave
- Thomas Scialom
- Jianfeng Gao
- Timo Schick
- Roberta Raileanu
In Natural Language Processing
- Timo Schick
- Asli Celikyilmaz
- Thomas Scialom
- Hinrich Schütze
- Edouard Grave
- Maria Lomeli
- Yejin Choi
- Jane Dwivedi-Yu
Citations in the corpus
- Augmented Language Models: a Survey (2023, methodology): “For the latter, examples of reasoning are typically used in the larger context of instruction tuning (Mishra et al., 2021; Sanh et al., 2022; Wang et al., 2022d; Ouyang et al., 2022), where, more generally, an LM’s ability to follow instructions is improved based on human-labeled examples.”