README does not contain install instructions

improve_utils,py needs to be present in the repo.

my_data_dir not present in the working directory.
It will be better to add mkdir my_data_dir in the ReadMe.

Preparing model repository prior to release

Documentation:

• Install instructions (conda env and Singularity definition file)

Model scripts:

• preprocess.py (works with benchmark data sets)
• train.py (takes output from preprocess.py)
• infer.py (takes model from train.py)
• tested on Benchmark data sets
• supports candle_lib checkpointing (LINK)

Create or update Interface scripts:

• preprocess.sh
• train.sh
• infer.sh
• Test scripts

Container:

• Create Singularity definition file
• Build and test container with improve branch

Code ready:

• Ready for release testing / hand over

Release

Version: v0.3.0-beta

• Build and test container with improve branch
• Tag for release
• Build and test container with release tag

SOP:

1. Setup/Install instructions
2. Interface scripts exist
3. Tested interface scripts locally
4. Singularity definition file
5. Create improve branch in the model repo
6. Switch to improve branch in image build
7. Test container build
8. Test train with container
9. Test HPO in Hall of Fame with checkpointing
10. Tag improve branch

SOP:

• Create improve branch in model repo
• Switch to improve branch in image build
• Test container build
• Test train with container
• Test HPO
• Tag improve branch with v0.6.0

1. singularity build --fakeroot build/HiDRA.sif definitions/HiDRA.def
2. singularity run --nv --bind pwd/tmp:/candle_data_dir build/HiDRA.sif train.sh 1 /candle_data_dir --epochs 1

Output:

  File "pandas/_libs/parsers.pyx", line 382, in pandas._libs.parsers.TextReader.__cinit__
  File "pandas/_libs/parsers.pyx", line 674, in pandas._libs.parsers.TextReader._setup_parser_source
FileNotFoundError: [Errno 2] No such file or directory: '/candle_data_dir/ge_GDSC1000.csv'