! This is a breaking update !
What breaks it, there is a change to the function signature of the train_model function. I take out the pretrain_rmse_weight and finetune_rmse_weight arguments. I also change the model argument name to model_type.
I hesitate to make the merge, but I also want to keep moving forward and I worry a little that if I just start going off on this branch, I will get out of sync so much with the master.
merging in grad_similarity branch.
! This is a breaking update ! What breaks it, there is a change to the function signature of the
train_model
function. I take out thepretrain_rmse_weight
andfinetune_rmse_weight
arguments. I also change themodel
argument name tomodel_type
.I hesitate to make the merge, but I also want to keep moving forward and I worry a little that if I just start going off on this branch, I will get out of sync so much with the master.