The code provides a simple implementation of behaviour cloning for the various environments in MineRL (in my case, MineRLTreechop-v0).
I have created a custom dataloader (data_pipeline.py and buffered_batch_iter.py) to allow support with MineRL v1.0 as well. However, due to issues with env.render() This code currently does not support MineRL v1.0 yet.
To be able to run this code, follow the installation steps for MineRL v0.4.4. Also install the necessary dataset from MineRL from here.
Uncomment train() in main if you wish to train the model yourself.
Run behavioural_cloning.py for training your model and testing it afterwards.
I was able to write this code with the help of a useful template provided by minerllabs.