I do not remember in great detail anymore what the exact problem is, but currently we need to patch CGLM pipelines since they do not have a start_timestamp set. If we set the start timestamp, we get many empty triggers. However, currently, without the start timestamp, the currently active/trained models are not set correctly. We should document more issues on what the underlying problem is and find a solution, because it's unintutitive that a pipeline file needs patching
I do not remember in great detail anymore what the exact problem is, but currently we need to patch CGLM pipelines since they do not have a
start_timestamp
set. If we set the start timestamp, we get many empty triggers. However, currently, without the start timestamp, the currently active/trained models are not set correctly. We should document more issues on what the underlying problem is and find a solution, because it's unintutitive that a pipeline file needs patching