solver_stepsize_6400.prototxt

net: "models/jhlim_googlenet_bn/train_val.prototxt"
#test_iter: 1000
#test_interval: 4000
#test_initialization: false
display: 40
average_loss: 40
base_lr: 0.045
lr_policy: "step"
stepsize: 6400
gamma: 0.96
max_iter: 1200000
momentum: 0.9
weight_decay: 0.0002
snapshot: 10000
snapshot_prefix: "models/jhlim_googlenet_bn/googlenet_bn_stepsize_6400"
solver_mode: GPU