Open youngzw opened 2 years ago
I modified lr=1e-3 (1e-4 in your paper) and found that the convergence speed is much faster. Similar phenomena also appeared in other code reproduction repositories, e.g, cr_1 and cr_2.
lr=1e-3
1e-4
I modified
lr=1e-3(1e-4in your paper) and found that the convergence speed is much faster. Similar phenomena also appeared in other code reproduction repositories, e.g, cr_1 and cr_2.