Deep Reinforcement Learning Implementations (Course Projects)
Published:
- Imitation Learning: Implemented Flow Matching vs. MSE for Action Chunking in Push-T environments.
- Online RL: Implemented major RL algorithms (e.g. GAE, DQN, SAC) on Atari and MuJoCo tasks.
- LLM RL: Fine-tuned LLMs using GRPO and GR-REINFORCE for complex mathematical reasoning tasks.
- Authored a literature review on In-context RL. [PDF]