A Comparative Study of DQN Training Methods on the Variable-Length CartPole Problem

This project provides implementations and reproducible experiments for assessing how different DQN training strategies generalize in environments with changing physical parameters. It includes Baseline DQN, Double DQN, and Curriculum Learning agents, all training scripts, fixed hyperparameters, test evaluation code, statistical tests (Friedman, Wilcoxon, Holm), and per-length results across 30 pole configurations.