PS-006-ML5G-PHY-Reinforcement-learning_IITI-RL

This repository is copy of challenge's main repo and the code is modified by following the challenge's rules and guidelines.

Team Members: Sundesh Gupta, Mohit Mehta, Bhagyashree Gour, Arijit Datta

Submission

Presentation and Videos (for best return and worst return respectively) are available at following link.

Reproducing Results

Curiosity model is trained using the script curiosity.ipynb and the model weights is saved in the directory model_state_dict.

To reproduce the leaderboard results, please run curiosity_eval.ipynb.

Baselines

The following are the baselines we tried:

1. DQN: RL.ipynb and RL_fed.ipynb
2. Advanced Actor Critic
3. Policy Gradient Network

References

1. https://github.com/jcwleo/curiosity-driven-exploration-pytorch
2. https://github.com/pytorch/examples/blob/master/reinforcement_learning/actor_critic.py
3. https://pytorch.org/tutorials/intermediate/reinforcement_q_learning.html
4. https://tims457.medium.com/policy-gradient-reinforcement-learning-in-pytorch-df1383ea0baf