Closed SirRob1997 closed 4 years ago
Took a shot at fixing this. Let me know if you have thoughts.
Once this has been fixed it is probably a good idea to run a new sweep. From that point onwards adding a new algorithm should only require running the new algorithm on all datasets since sampled hparams stay consistent for the others.
Just opening this here to keep track of the issue mentioned in #12:
"We will have to find a solution to guarantee that the same random parameters are sampled in
hparams_registry
given the same random seed, for all future commits. Right now, adding algorithms or conditional logic inhparams_registry
changes the random hyperparameters sampled for a given algorithm, dataset, and seed."