Self-Refined Large Language Model as Automated Reward Function Designer for Deep Reinforcement Learning in Robotics.
Jiayang SongZhehua ZhouJiawei LiuChunrong FangZhan ShuLei MaPublished in: CoRR (2023)
Keyphrases
- language model
- reward function
- reinforcement learning
- reinforcement learning algorithms
- markov decision processes
- language modeling
- state space
- context sensitive
- probabilistic model
- n gram
- optimal policy
- information retrieval
- function approximation
- inverse reinforcement learning
- multiple agents
- retrieval model
- transition model
- ad hoc information retrieval
- query expansion
- vector space model
- transition probabilities
- translation model
- query terms
- generative model
- markov chain
- learning algorithm
- machine learning
- mixture model
- initially unknown
- smoothing methods
- multi agent
- state variables
- document representation
- dynamic systems
- dynamic programming
- temporal difference
- model free
- graphical models
- image segmentation