REINFORCEOptions<T>
Options & Configuration · AiDotNet.Models.Options
Configuration options for REINFORCE agents.
Inheritance: ModelOptions → REINFORCEOptions<T>
For Beginners
REINFORCE is the “hello world” of policy gradient methods. It’s simple but powerful:
- Play an entire episode
- See which actions led to good rewards
- Make those actions more likely in the future
Think of it like learning to play a game: you play a round, see your score, then adjust your strategy to do better next time.
Simple, but can be slow to learn and high variance. Modern algorithms like PPO improve on REINFORCE’s ideas.
How It Works
REINFORCE is the simplest policy gradient algorithm. It directly optimizes the policy by following the gradient of expected returns.
Constructors
| Constructor | Summary |
|---|---|
REINFORCEOptions() |
Creates options with the default values. |
Properties
| Property | Summary |
|---|---|
List<int> HiddenLayers { get; set; } |
Hidden layer sizes for the policy network. |
T DiscountFactor { get; set; } |
Discount factor (gamma) for future rewards. |
T LearningRate { get; set; } |
Learning rate for policy gradient updates. |
bool IsContinuous { get; set; } |
Whether the action space is continuous (true) or discrete (false). |
int ActionSize { get; set; } |
Number of discrete actions available to the agent. |
int StateSize { get; set; } |
Dimension of the environment state vector. |