Simulate collaborative ML scenarios, experiment multi-partner learning approaches and measure respective contributions of different datasets to model performance.
add an option on how compute val score and test score. If set to local, the global test/val set will be split between partners exactly like the train set. it will also be corrupted as the train set is.
Then the val score and test score for global model will be the average of the scores computed on each partners set.
add an option on how compute val score and test score. If set to local, the global test/val set will be split between partners exactly like the train set. it will also be corrupted as the train set is.
Then the val score and test score for global model will be the average of the scores computed on each partners set.
closing #287