DRLC: Reinforcement Learning with Dense Rewards from LLM Critic.
Meng CaoLei ShuLei YuYun ZhuNevan WichersYinxiao LiuLei MengPublished in: CoRR (2024)
Keyphrases
- reinforcement learning
- actor critic
- function approximation
- temporal difference
- reinforcement learning algorithms
- markov decision processes
- state space
- reinforcement learning methods
- model free
- policy gradient
- learning algorithm
- reward function
- optimal policy
- transfer learning
- learning process
- robotic control
- total reward
- approximate dynamic programming
- multi agent
- rl algorithms
- optimal control
- reward shaping
- policy search
- temporal difference learning
- control policy
- stereo correspondence
- dynamic programming
- hidden state
- function approximators
- machine learning
- reinforcement learning problems
- monte carlo
- learning problems
- natural actor critic