Basic Policy Gradient Approach
A basic policy gradient approach to reinforcement learning involves three main steps. First, sample a number of state-action sequences (trajectories) based on a given policy. Second, evaluate each sampled sequence using a performance function that measures the expected cumulative reward. Third, update the model parameters to maximize this performance function, typically by employing optimization algorithms such as gradient descent.
0
1
Tags
Foundations of Large Language Models
Ch.4 Alignment - Foundations of Large Language Models
Foundations of Large Language Models Course
Computing Sciences
Related
Optimal Policy Parameters via Maximization Formula
An engineer is training a system using a reinforcement learning approach. The system's behavior is determined by a set of adjustable parameters. The training process aims to find the parameter values that maximize a specific 'performance function,' which represents the expected cumulative reward. The engineer runs two separate training procedures, Procedure X and Procedure Y, and observes the following final outcomes:
- Procedure X: The final set of parameters results in a performance function value of 150.
- Procedure Y: The final set of parameters results in a performance function value of 125. However, Procedure Y completed in half the time of Procedure X.
Which statement best evaluates the outcomes in relation to the primary training objective?
Evaluating Policy Effectiveness
Identifying Optimal Policy Parameters from Training Data
Basic Policy Gradient Approach