[FEATURE] - Run Arrakis from YAML files.

[yash-srivastava19]

One of the ways in which we can reduce experimentation time is if we can run arrakis using YAML files. We can specify what tools we can use in this file, and then run experiments from the CLI.

We can have the YAML file as :

field
  key1: val1
  key2: val2
...

And specifying the path to YAML file, and then parsing it, we can specify what tool we need to use. Another major enhancement we can add to it is to specify a base folder in which there are multiple "groups" of YAML files which we can provide as options as CLI arguments, this is similar to Hydra Multirun, where we can run experiments with different configs. Say, for example, the file we have a group1 with multiple YAML files as :

base
|___group1
      |____ optn1.yaml
      |____ optn2.yaml
...

Where optn1.yaml and optn2.yaml has same fields in it, but different values. Now, to run experiments we can just do a single command as

$ python -m multirun_test.py --multirun group1=optn1,optn2 
# in future, this will be changed to arrakis-cli --multirun group1=optn1,optn2 

And the experiment is run with both optn1.yaml and optn2.yaml.

We can also have multiple groups as :

$ python -m multirun_test.py --multirun group1=optn1,optn2 group2=optn1,optn2

And the experiment runs for - ((group1 option1) with (group2 option1)), ((group1 option1) with (group2 option2)) and other combinations. This is exactly what this feature wants to achieve. PRs welcome, and comment if you want to have a discussion on this. A minimal version, (although not completly correct) is in the future branch in the future folder. The template is similar to the one mentioned in this issue