You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
# @package data
defaults:
- dset_config_group: option1
task1 is just a logical category to keep all configurations related to one section of my experimental setting in the same place (I can later create another subpkg1, representing another section). I don't want task1 to be included in the hydra config object, therefore I override the package of dset.yaml to data only.
In addition, I would like to be able to select one between the two options in the config group. I can do this from the command line.
This setting however complicates the override in config.yaml, which now requires its own package override in the defaults. Now as a CLI arg I have to specify e.g. data/task1/dset_config_group@data.dset_config_group=option1. I feel this is a bit ugly and unnecessarily verbose in most cases. I would have preferred to just type data/task1/dset_config_group=option1, or data.dset_config_group=option1.
Questions:
Should my folder tree reflect only the structure of the cfg? In other words, should the folders be used only as config groups and not for any other kind of categorization?
Should I avoid CLI argument override in such cases, and rely more on config files for each experiment?
What is the reason behind defining such a verbose override syntax?
Is there a better way (in terms of folder organization or override syntax) to achieve what I want to achieve?
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
I have the following folder tree
the content of the important files is the following
config.yaml
dset.yaml
task1
is just a logical category to keep all configurations related to one section of my experimental setting in the same place (I can later create another subpkg1, representing another section). I don't wanttask1
to be included in the hydra config object, therefore I override the package ofdset.yaml
todata
only.In addition, I would like to be able to select one between the two options in the config group. I can do this from the command line.
This setting however complicates the override in config.yaml, which now requires its own package override in the defaults. Now as a CLI arg I have to specify e.g.
data/task1/dset_config_group@data.dset_config_group=option1
. I feel this is a bit ugly and unnecessarily verbose in most cases. I would have preferred to just typedata/task1/dset_config_group=option1
, ordata.dset_config_group=option1
.Questions:
cfg
? In other words, should the folders be used only as config groups and not for any other kind of categorization?Beta Was this translation helpful? Give feedback.
All reactions