RL-GPT: Integrating Reinforcement Learning and Code-as-policy.
Shaoteng LiuHaoqi YuanMinda HuYanwei LiYukang ChenShu LiuZongqing LuJiaya JiaPublished in: CoRR (2024)
Keyphrases
- reinforcement learning
- optimal policy
- rl algorithms
- policy search
- action selection
- actor critic
- reinforcement learning algorithms
- markov decision process
- markov decision processes
- reinforcement learning problems
- sequential decision making
- policy evaluation
- function approximation
- action space
- model free
- policy gradient
- control policies
- policy iteration
- state space
- markov decision problems
- reward function
- state and action spaces
- partially observable
- temporal difference
- function approximators
- control policy
- average reward
- decision problems
- partially observable domains
- state action
- continuous state
- total reward
- learning algorithm
- multi agent
- model free reinforcement learning
- control problems
- long run
- approximate dynamic programming
- machine learning
- partially observable environments
- optimal control
- partially observable markov decision processes
- approximate policy iteration
- exploration exploitation tradeoff
- learning process
- infinite horizon
- reinforcement learning methods
- finite state
- transfer learning
- transition model
- source code
- learning agents
- dynamic programming
- continuous state and action spaces
- temporal difference methods
- stochastic games
- autonomous learning
- direct policy search
- reward signal
- temporal difference learning
- average cost
- learning capabilities
- agent learns
- inverse reinforcement learning
- continuous state spaces