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
Each experiment will define a specific set of parameters including a specific dataset and a specific model.
I would like to call python train.py experiment=exp2 and automatically select the specified dataset and model and load their params. Is there a way of doing that? In the documentation it has Specializing configuration, but it only shows how to change a parameters instead of the whole config group.
If it is not possible, what would be the recommended alternative workflow?
The text was updated successfully, but these errors were encountered:
Hi, thanks for the question.
Recursive defaults list (#171) will be the solution for this and many other related problems.
This is not supported in Hydra 1.0 but is coming in 1.1 (Under active development in #1044).
In 1.0, you can only define the defaults list in your primary config.
Your alternatives are for now:
Use command line overrides to override the config group.
Use --config-name to override the primary config to one that has the defaults list you want.
There could be other alternatives I am not thinking of right now.
Hi, I am working in a project with configs in the following layout:
Each experiment will define a specific set of parameters including a specific dataset and a specific model.
I would like to call
python train.py experiment=exp2
and automatically select the specified dataset and model and load their params. Is there a way of doing that? In the documentation it has Specializing configuration, but it only shows how to change a parameters instead of the whole config group.If it is not possible, what would be the recommended alternative workflow?
The text was updated successfully, but these errors were encountered: