Task

A task in our context refers to a graph problem. The abstraction of such a problem enables us to experiment with multiple issues and understand the efficacy of our policy in varying circumstances.

Methods

Several methods are employed to set up the task, determine the reward, verify task completion, and evaluate node performance.

Setup

This step involves defining the initial action, observations, and the starting node.

Reward

Based on the new state, we determine the reward. This measure encourages the policy to make beneficial decisions.

Terminated

This method checks whether the task has been solved. It helps us understand the progress and efficiency of the policy.

Node Correctness

This method returns a color grade to visually represent how well a node has performed in its actions.

List of Tasks

Below are some tasks of special interest that we use to test our policy.

Coloring

(To Be Described)

Distance

(To Be Described)

The description of each task will provide a clear understanding of the problem context and the expected outcomes. This documentation will be updated as soon as the task descriptions are finalized.

TODO