There are the implementation of two algorithm for continuous learning, the Tile-Coding and the DQN for two problems from https://www.gymlibrary.dev. We used Sarsa for policy selection. The report explained the details and a code from the primary steps of learning is provided too.

First problem is the Mountain-Car and the second is Lunar-Lander.

The plot below is a comparison between different configurations for the Mountain-Car problem.

More details are provided in the report.