Juan M. Cruz-Martinez
Juan M. Cruz-Martinez
Due to the way the new commondata is implemented (grouped by paper), the suggestions (that looks at the entries in the folder) when a name is incorrect don't work anymore....
Before 4.0.9 theories were very big and fitting or loading them was very slow, and so "low precision" theories were used for the tests. This is no longer needed (as...
When running a hyperparameter scan, the memory grows proportional to the number of trials. This smells like a memory leak, since from one trial to the next virtually all information...
While in the server the `ekos` and the `fktables` are now separated, because it was easier to ensure that people didn't have to redownload theories, we left the ekos inside...
At the moment it just uses https://github.com/qiboteam/qibolab/pull/1138 to generate the random numbers which then are casted to tensorflow, which at that point is used basically as numpy. As it is...
This PR will implement the multiarchitecture parlallel training. The way I'm planning to do it is by first refactoring everything that happens below `generate_pdf_model`, which is the only function that...
The only reason it wasn't at this point was that the pseudodata was not written down. This is not a concern anymore.
At the moment we can only either drop all cuts (adding new `filter_rules`) or append cuts (with `added_filter_rules`). This PR adds `drop_internal_rules: [dataset list]` to drop all rules for a...
Closes #2151 Before merging I'll redo every test fit and check that every example runs, but first I'd like #2300 to be merged.
At the moment, due to implementation details at the time, the nuclear uncertainties (which are a theory contribution) are part of the experimental uncertianties. Ideally, the nuclear uncertainties should be...