About this document
Enhancing Autotelic Agents with LLMs by Sarath Chandar is a document available to read on EtoBox.
1. The document introduces an approach called LMA3 that augments autotelic agents with pretrained language models to support the representation, generation and learning of diverse goals. 2. LMA3 uses the language model as an imperfect model of human cultural transmission to help with three key components of autotelic agents: describing achieved goals, suggesting new goals, and defining reward functions for goals. 3. Without hand-coded representations, LMA3 agents are shown to learn a large diversity of s
- Author
- Sarath Chandar
- Language
- EN