James Owers

`# dtype = torch.device("cuda:0") # Uncomment this to run on GPU` should be `# device = torch.device("cuda:0") # Uncomment this to run on GPU`

The line referencing `self.load_vectors` https://github.com/codertimo/BERT-pytorch/blob/7efd2b5a631f18ebc83cd16886b8c6ee77a40750/bert_pytorch/dataset/vocab.py#L63 will fail as this method doesn't appear to exist (on quick search of the repo). Apologies if I've misunderstood something obvious...

We also might want to consider moving some of the functions from measure_errors into the mdtk package.

1

from #154 I agree, since there are useful functions contained within.

Automate documentation creation

2

We want to solve 3 problems regarding documentation: 1. We don't want to have to rerun the `ipynb` files in `./docs` after each change **by hand**, this is an issue...

Before release: DQA of the released data!

4

We should do some data quality analysis of the data we are going to release. I'm thinking a notebook (also doubles as an intro to what data are available for...

May have listed these elsewhere too, but noting as I go through docs * we should probably never create pianorolls prior to feeding the data to the model, but rather...

Implement use of logging as opposed to warnings

1

fix #88 then turn all warnings on and fix any problems that arise address #114

We're at 60%. To regenerate the report with lines that need covering, run `pytest -vv --cov --cov-report term-missing` Also you can get nice reports with: ```bash pytest --cov-report html open...

I didn't have time to do everything I wanted for travis. Things to add: * [ ] Make sure black, isort, and flake8 are run * [ ] Test all...