search

airoldilab / ai-sgd

Towards stability and optimality in stochastic gradient descent
5 stars 2 forks source link

Move the time benchmark function outside of exp_normal_correlated.R and make it generic #10

Open dustinvtran opened 9 years ago

dustinvtran commented 9 years ago

2ce70017f7a199403be9ac55023a34522350a67b makes progress on this. It still only benchmarks a fixed set of methods and a fixed data generating process (normal with equally correlated covariates).

dustinvtran commented 9 years ago

ece75ccd4359eb19a71419ba346fad5fe5b3b277 allows for a choice of methods.