param dim_i, dim_o loss opti dim_sample bwd_σ value dim_i=20 dim_o=30 mse mse+cos mse,cos all adam, lr=1e-5 adam, lr=1e-4 adam, lr=1e-3 adam, lr=1e-2 all 2^0 2^-2 2^-4 2^-6 2^-8 all loglinexp all hide confusion show confusion hide ext_loss show ext_loss hide loss_mse show loss_mse hide loss_cos show loss_cos hide loss_dot show loss_dot hide loss_dot_sim show loss_dot_sim