I have the following use case. I am working on a large project with multiple subtasks. For example, we are performing classification on some datasets, segmentation on others, and translation on yet another set of datasets (see below).

Each task has its own configuration. However, there is a part of the configuration that is shared among all tasks (such as the architecture of some network or specific blocks). This shared configuration is evolving, and each time I make a small change to it, I need to copy and paste the updated configuration into all the relevant folders.

Is there a way to centralize this configuration so that it can be imported into all the subtasks?

my_project
| classification_task
|--| conf
|--|--| config.yaml
|--|--| models
|--|--| datasets
|--| train.py
|--| ...
| segmentation_task
|--| conf
|--|--| config.yaml
|--|--| models
|--|--| datasets
|--| train.py
|--| ...
| translation_task
|--| conf
|--|--| config.yaml
|--|--| models
|--|--| datasets
|--| train.py
|--| ...