This repository collects my hands-on implementations across multiple reinforcement learning topics. The goal is to build practical, well-structured baselines and learn how different RL methods behave in real environments.
This project is organized by subject area, including:
- Imitation Learning
- Q-Learning
- Other value-based, policy-based, and model-free RL experiments
Imitation learning is one major part of this repository. In this module, the agent learns from expert demonstrations rather than learning only from reward-based exploration.
Implemented components include:
- Expert demonstration loading and preprocessing
- Behavior cloning for supervised policy learning
- Trajectory collection and rollout utilities
- Replay buffer sampling for training
- Training, evaluation, and logging pipeline
The long-term aim is to keep this repository modular so each RL topic can be studied, compared, and extended independently (for example, comparing imitation learning with Q-learning methods under similar settings).