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
Right now performance (up to DL3) is stored under protopipe/docs/contribute/benchmarks with DL3 being uploaded here (only poster plot though)
Of course, with more than 1 analysis this becomes not ideal...
What to expect around v0.5.0
I am also testing a nice way to show all benchmarking using Jupyter Book for which I will most likely open a PR soon (just to allow for a first raw solution).
In general, I would say that it won't be ideal to store all the benchmarking material (the executed notebooks) under the documentation in the future (I will most likely do it in v0.5.0, but I hope in the future we choose a different solution).
A possible solution for the future
A possible solution would be to open a repo only for the storage of performances (maybe already on under the DPPS Gitlab project?).
This repo would have a structure like so:
- protopipe version
- Production
- Site
- NSB level
- Zenith angle
- Azimuth direction
- source type
Inside each folder, there would be
the material required to showcase an HTML version of the book (a folder containing the benchmarking notebooks and the book material)
all the necessary to retrieve/reproduce the results (a Docker recipe, an analysis metadata file produced by the interface or similar).
This repository would have it's own "documentation" in which we link to the different books/analyses with external links to get the data or reproduce it (actually after looking a bit more into it, I think it would make sense to make the repo itself a book and each part is 1 analysis...but we'll see).
The text was updated successfully, but these errors were encountered:
UPDATE
Current situation
Right now performance (up to DL3) is stored under
protopipe/docs/contribute/benchmarks
with DL3 being uploaded here (only poster plot though)Of course, with more than 1 analysis this becomes not ideal...
What to expect around v0.5.0
I am also testing a nice way to show all benchmarking using Jupyter Book for which I will most likely open a PR soon (just to allow for a first raw solution).
In general, I would say that it won't be ideal to store all the benchmarking material (the executed notebooks) under the documentation in the future (I will most likely do it in v0.5.0, but I hope in the future we choose a different solution).
A possible solution for the future
A possible solution would be to open a repo only for the storage of performances (maybe already on under the DPPS Gitlab project?).
This repo would have a structure like so:
Inside each folder, there would be
This repository would have it's own "documentation" in which we link to the different books/analyses with external links to get the data or reproduce it (actually after looking a bit more into it, I think it would make sense to make the repo itself a book and each part is 1 analysis...but we'll see).
The text was updated successfully, but these errors were encountered: