Inspiration
We wanted to solve the game presented by Capchase with a Deep Reinforcement algorithm.
What it does
We have a notebook that executes some games played by our agent, which "learns" a strategy to win the game.
How we built it
We used Python and a Kaggle notebook. Pytorch and some other standard python libs. Then we wanted to bring the learned parameters into Rhai as a hardcoded vector and use some self implemented functions to run the network in inference time.
Challenges we ran into
The algorithm didn't learn in the python framework. Also, as the code to submit must be in Rhai (a very simple framework) we needed to implement a lot of calculus in order to be able to run our model. At the end, we weren't able to submit the code.
Accomplishments that we're proud of
We developed a running code!
What we learned
Some members of the team learned the Q learning algorithm (some others aready knew it) and we all learnt Rhai.
Built With
- python
- rhai
Log in or sign up for Devpost to join the conversation.